Cloud-Computing-–II-munotes

Page 1

1 1
PARALLEL AND DISTRIBUTED
COMPUTING
Unit Structure :
1.0 Objectives
1.1 Introduction
1.2 Parallel And Distributed Computing
1.3 Elements of parallel computing, elements of distributed computing
1.4 Technologies for distributed computing
1.5 RPC
1.6 Distributed object frameworks
1.7 Service oriented computing
1.8 Virtualization
1.9 Summary
1.10 References
1.11 Unit End Question
1.0 OBJECTIVES
After learning this chapter, you will be able to understand different cloud
computing technologies avai lable. It will help you in better understanding
of parallel and distributed framework.
1.1 INTRODUCTION
Cloud computing is a revolutionary paradigm that has transformed the
way businesses and individuals’ access and utilize computing resources. It
refers t o the delivery of on -demand computing services over the internet,
allowing users to store, process, and access data and applications without
the need for local infrastructure.
At its core, cloud computing provides a flexible and scalable model for
resource allocation. Instead of relying on local servers and data centers,
users can leverage a vast network of remote servers maintained by cloud
service providers. This eliminates the need for costly hardware
investments and the burden of managing and maintainin g infrastructure.
munotes.in

Page 2

Cloud Computing –II
2 One of the key advantages of cloud computing is its elasticity. Users can
easily scale their resources up or down based on demand, ensuring optimal
performance and cost -efficiency. Whether it's a small startup requiring
minimal resources or a large enterprise with heavy computational needs,
cloud computing allows for seamless scalability.
Cloud computing offers several service models to cater to different
requirements. Infrastructure as a Service (IaaS) provides virtualized
computing reso urces such as virtual machines, storage, and networks.
Platform as a Service (PaaS) offers a platform on which developers can
build and deploy applications, while Software as a Service (SaaS)
provides ready -to-use software applications accessible through t he cloud.
Moreover, cloud computing facilitates collaboration and remote access.
Users can access their data and applications from any device with an
internet connection, enabling remote work and enhancing productivity. It
also fosters data sharing and col laboration among teams by providing a
centralized platform accessible to authorized users.
Security is a crucial aspect of cloud computing. Cloud service providers
employ robust security measures to protect data from unauthorized access,
ensuring data priv acy and integrity. They also implement backup and
disaster recovery mechanisms to safeguard against data loss.
The adoption of cloud computing has had a profound impact on various
industries. It has enabled organizations to focus on their core
competencies while offloading the responsibility of managing
infrastructure. It has empowered startups with cost -effective solutions and
leveled the playing field for businesses of all sizes.
In conclusion, cloud computing has revolutionized the IT landscape by
provid ing scalable, on -demand computing resources. Its flexibility,
accessibility, and cost -efficiency make it an attractive choice for
individuals and businesses alike. As technology advances further, cloud
computing is likely to continue its rapid growth, shap ing the future of
computing and transforming the way we live and work.
1.2 PARALLEL AND DISTRIBUTED COMPUTING
Parallel and distributed computing are two related concepts in the field of
computer science that aim to increase computational power and efficien cy
by dividing tasks among multiple processors or machines. While they
share some similarities, they differ in terms of architecture and resource
management. Let's explore parallel and distributed computing in more
detail.
Parallel computing involves the s imultaneous execution of multiple
computational tasks to solve a larger problem. It leverages the power of
multiple processors or cores within a single machine, working together to
complete a task faster. The processors communicate and collaborate with
each other to divide the workload and share data. This allows for efficient munotes.in

Page 3

Parallel and Distributed Computing
3 utilization of computing resources and accelerates the overall
computation.
In parallel computing, shared memory and shared address space are
commonly used. Shared memory allows proce ssors to access a common
memory location, enabling efficient data sharing and communication.
Each processor operates independently on a portion of the problem, and
synchronization mechanisms ensure that processors coordinate their
actions and avoid conflic ts.
The diagram below illustrates a basic parallel computing architecture with
multiple processors connected to a shared memory system:
_____________________________
| Processor 1 |
|_____________________________|
| Proce ssor 2 |
|_____________________________|
| Processor 3 |
|_____________________________|
| Shared Memory |
|_____________________________|
Distributed computing, on the other hand, involves the execution of a
computational task across multiple interconnected machines or nodes.
Each node in a distributed system has its own memory and processing
unit, and the nodes communicate and coordinate their actions to solve a
problem collectively. Distributed computing is often used to solve
problems that require a vast amount of computational power, storage, or
data processing capabilities.
In distributed computing, message passing is a common communication
mechanism. Nodes exchange messages to share data and coordinate their
activities. Distributed systems employ various algorithms and protocols to
manage tasks distribution, load balancing, fault tolerance, and data
consistency across multiple nodes.
The diagram below illustrates a basic distributed computing architectu re
with interconnected nodes:
________ ________ ________
| Node 1 | | Node 2 | | Node 3 |
|________| |________| |________|
|____________|____________|
Network munotes.in

Page 4

Cloud Computing –II
4 In a distributed computing environment, e ach node operates
independently, working on its portion of the task. The nodes communicate
and share intermediate results to collectively solve the problem.
Distributed systems can be geographically distributed across multiple
locations, enabling efficient resource utilization and fault tolerance.

Parallel and distributed computing techniques are often combined to
achieve even higher levels of performance and scalability. For example, a
large -scale problem can be divided into smaller subproblems, where ea ch
subproblem is solved in parallel on a cluster of distributed machines. This
approach, known as parallel -distributed computing, allows for efficient
utilization of both parallel processing within each machine and distributed
computing across multiple mac hines.
1.3 ELEMENTS OF PARALLEL COMPUTING,
ELEMENTS OF DISTRIBUTED COMPUTING
Elements of Parallel Computing:
1. Task Decomposition: In parallel computing, a problem is divided into
smaller tasks that can be executed concurrently. Task decomposition
involv es identifying the dependencies and breaking down the problem
into smaller, independent subtasks that can be executed simultaneously
on different processors or cores.
2. Data Partitioning: Data partitioning involves dividing the data
associated with a problem among different processors or cores. Each
processor operates on its assigned data subset, minimizing data
movement and maximizing locality, which can improve performance.
3. Synchronization: Synchronization mechanisms are essential for
coordinating the activ ities of parallel processes. They ensure that
multiple processes access shared resources or data in a controlled
manner to avoid conflicts and maintain data integrity. Common
synchronization techniques include locks, barriers, semaphores, and
atomic operat ions.
4. Load Balancing: Load balancing aims to distribute the workload
evenly among processors or cores to maximize resource utilization and
minimize idle time. Load balancing techniques dynamically allocate
tasks or data chunks to processors based on their current workload or
performance characteristics. munotes.in

Page 5

Parallel and Distributed Computing
5 5. Communication: Communication refers to the exchange of data and
messages between parallel processes. Efficient communication
mechanisms are crucial for sharing data, coordinating activities, and
synchronizin g the execution of parallel tasks. Communication can
occur through shared memory or message passing.
Elements of Distributed Computing:
1. Resource Discovery: In distributed computing, resource discovery
involves locating and identifying available computing r esources (e.g.,
nodes, servers, services) within a distributed system. Resource
discovery mechanisms allow nodes to find and utilize resources
efficiently.
2. Communication Protocols: Distributed computing relies on
communication protocols for nodes to exchan ge information and
coordinate their activities. Protocols define the rules and formats for
data transfer, message passing, and synchronization across the
distributed system.
3. Fault Tolerance: Distributed systems are prone to failures, such as
node crashes o r network disruptions. Fault tolerance mechanisms
ensure system resilience and reliability by detecting and handling
failures. Techniques like replication, redundancy, and error detection
and recovery are employed to maintain system integrity.
4. Consistency and Data Management: Consistency refers to the
agreement or coherence of data across distributed nodes. Distributed
systems often deal with shared data, and ensuring data consistency is
crucial. Consistency protocols and distributed data management
techniq ues, like distributed databases or distributed file systems, are
used to maintain data integrity and coherence.
5. Scalability: Distributed computing emphasizes scalability to
accommodate the increasing demands of users and applications.
Scalability involves the ability to add or remove nodes dynamically,
handle growing workloads, and distribute resources efficiently to
ensure optimal performance.
Both parallel and distributed computing share elements such as
communication, synchronization, and scalability. Ho wever, they differ in
terms of the underlying architecture and resource management strategies.
Understanding these elements is essential for designing and developing
efficient parallel and distributed computing systems.
1.4 TECHNOLOGIES FOR DISTRIBUTED
COMPUTING
There are several technologies available for distributed computing that
enable the efficient execution of tasks across multiple interconnected
machines or nodes. These technologies provide frameworks, libraries, and
platforms for developing dist ributed applications and managing the munotes.in

Page 6

Cloud Computing –II
6 complexities of distributed systems. Here are some notable technologies
for distributed computing:
1. Apache Hadoop: Hadoop is a popular open -source framework for
distributed storage and processing of large -scale data set s. It consists
of the Hadoop Distributed File System (HDFS) for distributed storage
and the MapReduce programming model for distributed processing.
Hadoop enables parallel and fault -tolerant processing of big data by
distributing data and computation acros s a cluster of machines.
2. Apache Spark: Spark is an open -source distributed computing system
that provides a high -level API for large -scale data processing. It offers
in-memory computing, allowing for faster data processing compared
to disk -based systems li ke Hadoop. Spark supports various
programming languages, such as Scala, Java, Python, and R, and
provides libraries for machine learning, graph processing, and stream
processing.
3. Apache Kafka: Kafka is a distributed streaming platform that enables
the proc essing of real -time data streams. It provides high -throughput,
fault-tolerant, and scalable messaging capabilities, allowing data to be
reliably published, subscribed, and processed in real -time. Kafka is
commonly used in distributed architectures for buil ding event -driven
systems and stream processing applications.
4. Apache ZooKeeper: ZooKeeper is a distributed coordination service
that helps manage and synchronize distributed applications. It provides
primitives such as distributed locks, configuration mana gement, and
leader election, which are essential for maintaining consistency,
reliability, and fault tolerance in distributed systems. ZooKeeper is
widely used in distributed databases, messaging systems, and other
distributed applications.
5. Kubernetes: Kubernetes is an open -source container orchestration
platform that facilitates the management and scaling of containerized
applications across distributed clusters of machines. It automates tasks
such as deployment, scaling, and load balancing, ensuring effic ient
resource utilization and fault tolerance. Kubernetes is commonly used
for deploying and managing microservices -based distributed
applications.
6. Apache Cassandra: Cassandra is a distributed NoSQL database that
offers high scalability and fault tolerance . It is designed to handle large
amounts of data across multiple nodes and data centers, providing high
availability and low -latency access to data. Cassandra's decentralized
architecture and data replication mechanisms make it well -suited for
distributed applications requiring high performance and fault
tolerance.
7. Amazon Web Services (AWS) and Microsoft Azure: AWS and
Azure are cloud computing platforms that offer a wide range of
distributed computing services. These platforms provide scalable and munotes.in

Page 7

Parallel and Distributed Computing
7 managed infrastructure for distributed applications, including virtual
machines, container services, serverless computing, data storage,
messaging services, and more. They enable developers to build and
deploy distributed applications without the need for extensiv e
infrastructure management.
These are just a few examples of the many technologies available for
distributed computing. Other notable technologies include Apache Storm
for stream processing, Apache Flink for batch and stream processing,
RabbitMQ for distr ibuted messaging, and Google Cloud Platform (GCP)
services like BigQuery, Pub/Sub, and Dataflow. The choice of technology
depends on the specific requirements of the distributed application,
including data volume, processing speed, fault tolerance, and sca lability
needs.
1.5 RPC
RPC (Remote Procedure Call) is a communication protocol commonly
used in cloud computing environments to enable interaction between
different components or services distributed across a network. RPC allows
a program or service to in voke a procedure or function on a remote system
as if it were a local call, abstracting the complexities of network
communication.
In a cloud computing context, RPC facilitates the seamless communication
between cloud services, allowing them to interact an d exchange data
efficiently. Here's how RPC is utilized in the cloud:
Service Invocation: Cloud services can use RPC to invoke procedures or
functions on other services deployed in the cloud. This enables them to
utilize the functionalities and capabilitie s offered by different services. For
example, a storage service might use RPC to invoke a data processing
service to perform analytics on stored data.
1. Inter -Service Communication: RPC serves as a communication
mechanism for cloud services to exchange data and messages. Services
can use RPC to request and receive data from other services, enabling
them to collaborate and share information. This facilitates the
development of modular and scalable cloud architectures, where
services can communicate and interac t seamlessly.
2. Service Orchestration: In a cloud environment, RPC plays a vital
role in service orchestration. Orchestrating services involves
coordinating the execution of multiple services to achieve a specific
task or workflow. RPC allows services to com municate and
synchronize their activities, ensuring proper sequencing and
coordination. This enables the composition of complex workflows and
distributed systems in the cloud.
3. API Invocation: RPC is commonly used in cloud APIs (Application
Programming Inte rfaces) to enable developers to interact with cloud
services. Cloud providers expose APIs that allow developers to invoke munotes.in

Page 8

Cloud Computing –II
8 various operations and functionalities provided by the cloud platform.
RPC facilitates the communication between client applications a nd the
cloud API, allowing developers to integrate cloud services into their
applications seamlessly.
4. Client -Server Communication: RPC is employed in client -server
communication scenarios in the cloud. Clients can initiate RPC calls to
remote server -side c omponents or services to request data, perform
operations, or trigger specific actions. The RPC protocol handles the
serialization, transmission, and deserialization of data between the
client and server, providing a transparent and efficient communication
channel.
Overall, RPC plays a crucial role in enabling communication and
interaction between distributed components and services in the cloud. It
simplifies the development and integration of cloud services, allowing
them to work together seamlessly and e fficiently. By abstracting the
complexities of network communication, RPC helps developers focus on
building scalable and modular cloud applications that leverage the
capabilities offered by different cloud services.
1.6 DISTRIBUTED OBJECT FRAMEWORKS
Distr ibuted object frameworks are software frameworks that provide a
programming model and infrastructure for developing distributed
applications in cloud computing environments. These frameworks
simplify the development of distributed systems by abstracting th e
complexities of network communication, data sharing, and resource
management. They allow developers to work with distributed objects,
which are objects that span multiple machines or nodes in a networked
environment. Here are some notable distributed obj ect frameworks used in
the cloud:
1. Java RMI (Remote Method Invocation): Java RMI is a distributed
object framework in the Java ecosystem. It enables Java objects to
invoke methods on remote objects residing on different machines or
JVMs. RMI provides a tran sparent mechanism for remote method
invocation, where the developer can invoke methods on remote objects
as if they were local objects. It supports serialization, object activation,
and garbage collection across distributed systems.
2. CORBA (Common Object Re quest Broker Architecture): CORBA
is a distributed object framework that provides a standardized model
for building distributed systems. It supports multiple programming
languages and platforms, allowing objects written in different
languages to communicat e and interact seamlessly. CORBA uses an
Object Request Broker (ORB) to handle communication between
objects, providing transparent object invocation and location
transparency. munotes.in

Page 9

Parallel and Distributed Computing
9 3. Microsoft .NET Remoting: .NET Remoting is a distributed object
framework for de veloping distributed applications in the Microsoft
.NET ecosystem. It allows objects written in .NET languages to be
accessed and invoked remotely. .NET Remoting provides a flexible
and extensible programming model for developing distributed systems,
suppo rting various communication protocols and transport
mechanisms.
4. gRPC: gRPC is a modern open -source framework developed by
Google that facilitates the development of high -performance
distributed systems. It uses the Protocol Buffers data format for
efficien t serialization and supports multiple programming languages.
gRPC provides a flexible and extensible RPC -based communication
model and supports features like bidirectional streaming,
authentication, and load balancing. It is widely used in cloud -native
applications and microservices architectures.
5. Apache Thrift: Apache Thrift is an open -source distributed object
framework developed by Apache Software Foundation. It enables
efficient cross -language service development and supports a wide
range of programming languages. Thrift uses a binary protocol and
provides an interface definition language (IDL) for defining services
and data structures. It supports various transport protocols, including
HTTP, TCP, and named pipes.
6. Erlang OTP: Erlang OTP (Open Telecom Pla tform) is a concurrent
and distributed programming framework designed for building fault -
tolerant and highly available systems. OTP provides abstractions for
distributed computing, including distributed process management,
message passing, and fault recove ry. It is widely used in
telecommunications and messaging systems, where fault tolerance and
high availability are critical.
These distributed object frameworks help developers build scalable,
modular, and distributed applications in cloud computing enviro nments.
They handle the complexities of communication, data sharing, and
resource management, allowing developers to focus on application logic.
These frameworks provide features like location transparency, object
serialization, and remote method invocatio n, enabling developers to work
with distributed objects seamlessly. By abstracting the intricacies of
distributed systems, these frameworks promote code reuse, reduce
development effort, and facilitate the creation of distributed applications in
the cloud.
1.7 SERVICE ORIENTED COMPUTING
Service -Oriented Computing (SOC) is a computing paradigm that focuses
on designing and building software systems as a collection of loosely
coupled, interoperable services. In SOC, services are self -contained,
modular units of functionality that can be independently deployed,
invoked, and composed to fulfil specific business requirements. These munotes.in

Page 10

Cloud Computing –II
10 services communicate with each other through well -defined interfaces,
enabling flexibility, scalability, and reusability in software development.
Here's an overview of Service -Oriented Computing:
1. Service: A service is a self -contained unit of functionality that
encapsulates business logic and provides well -defined interfaces for
communication. Services can be implemented using various
technologies, such as web services, RESTful APIs, or microservices.
Services have clear boundaries, are platform -independent, and can be
developed using different programming languages and frameworks.
2. Service Description: A service is described using a serv ice
description language, such as Web Services Description Language
(WSDL) or OpenAPI, which specifies the service's interface,
operations, input/output parameters, and protocols for communication.
The service description enables service discovery and allo ws clients to
understand how to interact with the service.
3. Service Discovery: Service discovery mechanisms enable clients to
locate and identify available services in a distributed environment.
Service registries, such as UDDI (Universal Description, Disco very,
and Integration), or service discovery frameworks like Consul or
Eureka, provide a centralized repository or distributed mechanisms for
registering and discovering services. Clients can query the registry to
find services that match their requirement s and obtain their endpoint
information.
4. Service Composition: Service composition involves combining
multiple services to create more complex and higher -level business
functionalities. Services can be composed dynamically at runtime to
fulfill specific req uirements. Service composition can be achieved
through choreography, where services collaborate based on predefined
message exchanges, or orchestration, where a central controller
coordinates the execution of services to achieve a specific business
process .
5. Service Orchestration: Service orchestration involves designing and
managing the flow of services to achieve a specific business goal. An
orchestrator, typically a centralized component, coordinates the
execution of services by invoking their operations in a predefined
sequence. Orchestration allows for complex business processes to be
modeled and managed effectively by coordinating the interactions
between services.
6. Service Choreography: Service choreography refers to the
collaboration and coordination o f services without a central controller.
Services interact based on predefined message exchanges and event -
driven communication. Choreography enables more decentralized and
loosely coupled interactions between services, allowing for flexibility
and scalabi lity in complex systems. munotes.in

Page 11

Parallel and Distributed Computing
11 7. Service Governance: Service governance focuses on managing and
controlling the lifecycle of services within an organization or
ecosystem. It involves policies, standards, and guidelines for service
development, deployment, versioni ng, security, and quality assurance.
Service governance ensures consistency, compliance, and proper
management of services throughout their lifecycle.
8. Service Security: SOC emphasizes security in service interactions.
Service security mechanisms include au thentication, authorization,
encryption, and message integrity to ensure secure communication
between services. Security policies and mechanisms are implemented
at both service endpoints and in the underlying communication
infrastructure to protect against potential threats and vulnerabilities.
Service -Oriented Computing promotes modularity, interoperability, and
flexibility in software development. It enables organizations to build
scalable and adaptable systems by leveraging existing services, promoting
code reuse, and facilitating system integration. SOC provides a foundation
for building distributed architectures, cloud -native applications, and
microservices -based systems, supporting agility and the ability to respond
to changing business needs.
1.8 VIRT UALIZATION
Virtualization is a technology that enables the creation of virtual instances
or representations of computing resources, such as servers, operating
systems, storage devices, or networks. It allows multiple virtual
environments to run concurrent ly on a single physical infrastructure,
providing improved efficiency, flexibility, and resource utilization. Here's
an overview of virtualization:
1. Server Virtualization: Server virtualization is one of the most
common forms of virtualization. It involves partitioning a physical
server into multiple virtual machines (VMs), each running its own
operating system and applications. Server virtualization allows for the
consolidation of multiple servers onto a single physical machine,
reducing hardware costs, pow er consumption, and physical space
requirements.
2. Hypervisor: A hypervisor, also known as a virtual machine monitor
(VMM), is the software layer that enables the creation and
management of virtual machines. It abstracts the underlying hardware
and provides a virtualization layer that allows multiple VMs to run on
a single physical server. The hypervisor ensures isolation, resource
allocation, and facilitates the communication between VMs and the
physical hardware.
3. Operating System Virtualization: Operating s ystem (OS)
virtualization, also known as containerization, allows multiple isolated
user spaces or containers to run on a single operating system kernel.
Containers share the host operating system, libraries, and resources, munotes.in

Page 12

Cloud Computing –II
12 but provide separate and isolate d execution environments. OS
virtualization offers lightweight and fast deployment of applications,
enabling efficient resource utilization and scalability.
4. Desktop Virtualization: Desktop virtualization enables the creation
of virtual desktop infrastructu re (VDI) by hosting multiple desktop
environments on a centralized server or data center. Users can access
their virtual desktops remotely, allowing for flexibility and mobility.
Desktop virtualization simplifies desktop management, enhances
security, and enables remote collaboration.
5. Storage Virtualization: Storage virtualization abstracts physical
storage resources, such as hard drives or storage arrays, and presents
them as logical storage pools. It allows for centralized management,
improved scalability , and simplified storage provisioning. Storage
virtualization enables features like data migration, snapshots,
replication, and backup, enhancing data availability and disaster
recovery.
6. Network Virtualization: Network virtualization abstracts the physical
network infrastructure, creating virtual networks that operate
independently and are isolated from each other. Virtual networks can
be provisioned and managed programmatically, allowing for efficient
allocation of network resources and simplified network management.
Network virtualization enables the creation of virtual LANs, virtual
switches, and virtual routers, facilitating multi -tenancy, segmentation,
and network flexibility.
Benefits of Virtualization:
Virtualization offers several benefits, includin g:
 Increased Efficiency: Virtualization enables better utilization of
hardware resources, leading to cost savings, reduced power
consumption, and improved efficiency
 Flexibility and Scalability: Virtualization provides the ability to
dynamically allocate a nd adjust resources based on workload
demands, allowing for scalability and flexibility.
 Consolidation and Resource Optimization: Virtualization allows for
the consolidation of multiple systems onto a single physical
infrastructure, reducing hardware and m anagement costs.
 Improved Disaster Recovery and High Availability: Virtualization
enables the creation of snapshots, backups, and replicas, facilitating
faster disaster recovery and enhancing system availability.
 Simplified Management: Virtualization centr alizes management and
provisioning, making it easier to deploy, monitor, and maintain
virtualized environments.
munotes.in

Page 13

Parallel and Distributed Computing
13 Virtualization has become a fundamental technology in modern data
centers and cloud computing environments, enabling efficient resource
utiliza tion, agility, and cost savings. It forms the foundation for
technologies like cloud computing, containerization, and software -defined
infrastructure, revolutionizing the way computing resources are
provisioned, managed, and utilized.
1.9 SUMMARY
Cloud com puting is a paradigm that involves the delivery of on -demand
computing resources over the internet. It provides organizations and
individuals with access to a shared pool of configurable computing
resources, including networks, servers, storage, applicatio ns, and services,
that can be rapidly provisioned and released with minimal management
effort.
Cloud computing offers several advantages, such as:
1. Scalability: Cloud resources can be easily scaled up or down based on
demand. Organizations can quickly alloc ate additional resources
during peak periods and scale them back when demand decreases,
enabling cost optimization and efficient resource utilization.
2. Cost Efficiency: Cloud computing eliminates the need for upfront
infrastructure investments. Organization s can pay for cloud services on
a pay -as-you-go basis, allowing for cost savings by reducing capital
expenditure and shifting to operational expenditure.
3. Flexibility and Agility: Cloud services provide flexibility in terms of
access and usage. Users can ac cess cloud resources from anywhere
with an internet connection and can choose the type and level of
services based on their specific requirements. Cloud platforms also
offer rapid deployment of applications and services, enabling faster
time-to-market for businesses.
4. Reliability and High Availability: Cloud service providers typically
have robust infrastructure and redundant systems to ensure high
availability and reliability. They often provide service level
agreements (SLAs) that guarantee a certain level of uptime and
performance, minimizing downtime and ensuring business continuity.
5. Security: Cloud providers invest heavily in security measures to
protect customer data. They implement advanced security controls,
encryption, access management, and regular audits to ensure data
privacy and protection.
6. Collaboration and Scalable Services: Cloud computing enables
seamless collaboration by providing shared access to resources,
allowing teams to work on the same documents and applications
simultaneously. It also offers a wide range of pre -built services and
APIs that can be easily integrated into applications, enabling the
development of scalable and feature -rich solutions. munotes.in

Page 14

Cloud Computing –II
14 7. Disaster Recovery: Cloud computing facilitates efficient data backup,
replication, and dis aster recovery solutions. Organizations can
replicate their data and systems across multiple geographical locations,
ensuring data redundancy and minimizing the risk of data loss.
Cloud computing is categorized into different service models:
1. Infrastructure as a Service (IaaS): Provides virtualized computing
resources, such as virtual machines, storage, and networks, allowing
users to deploy and manage their applications and systems.
2. Platform as a Service (PaaS): Offers a platform and runtime
environment for developing, testing, and deploying applications
without the need to manage the underlying infrastructure. PaaS
provides a higher level of abstraction, enabling developers to focus on
application development rather than infrastructure management.
3. Software as a Service (SaaS): Delivers ready -to-use software
applications over the internet on a subscription basis. Users can access
and use these applications through a web browser or thin client
without the need for local installation or maintenance.
Cloud compu ting has revolutionized the IT industry, enabling
organizations to focus on their core business activities while leveraging
scalable and cost -effective computing resources. It has become an
essential component of digital transformation strategies, empoweri ng
businesses to innovate, collaborate, and scale their operations efficiently.
1.10 REFERENCES
Enterprise Cloud Computing Technology, Architecture, Applications,
Gautam Shroff, Cambridge University Press, 2010
 Mastering In Cloud Computing, Rajkumar Buyya , Christian Vecchiola
And ThamariSelvi S, Tata Mcgraw -Hill Education, 2013
 Cloud Computing: A Practical Approach, Anthony T Velte, Tata
Mcgraw Hill, 2009
References:
 Architecting the Cloud: Design Decisions for Cloud Computing
Service Models
 (SaaS, Paa S, and IaaS), Michael J. Kavis, Wiley CIO, 2014
 Cloud Computing: SaaS, PaaS, IaaS, Virtualization, Business Models,
Mobile,
Security and More, Kris Jamsa, Jones & Bartlett Learning, 2013

munotes.in

Page 15

Parallel and Distributed Computing
15 1.11 UNIT END QUESTION
1. Explain Parallel computing.
2. Explain Distri buted Computing.
3. Difference between Parallel and Distributed system.
4. Explain elements of parallel and distributed computing.
5. Explain Virtualization in detail
6. Explain different Technologies for distributed computing.


munotes.in

Page 16

16
2
COMPUTING PLATFORMS -I
Unit Structure
2.0 Objective
2.1 Introduction
2.2 Cloud Computing definition and characteristics,
2.3 Enterprise Computing,
2.4 The internet as a platform,
2.5 Summary
2.6 Reference for further reading
2.7 Unit End Exerc ises
2.0 OBJECTIVE
 To understand the concept of cloud computing technologies.
 To learn different types of virtualization in cloud computing.
 To understand the different characteristics of cloud computing.
 To understand the how and why the internet also evo lved into a
platform for enterprise applications
2.1 INTRODUCTION
Cloud computing technologies are:
 Virtualization
 Service -Oriented Architecture (SOA)
 Grid Computing
 Utility Computing
Virtualization
“Virtualization means the process of creating a virtual e nvironment to run
multiple applications and operating systems on the same server”.
munotes.in

Page 17

Computing Platforms -I
17 Types of Virtualization
1. Hardware virtualization
2. Server virtualization
3. Storage virtualization
4. Operating system virtualization
5. Data Virtualization
Service -Oriented Architectu re (SOA)
Service -Oriented Architecture enables organizations to access on -demand
cloud -based computing solutions on the report of the change of business
requirement.
Applications of Service -Oriented Architecture
1. It is used in the healthcare industry.
2. It is used to create different mobile applications and games using SOA.
3. In the air force, SOA infrastructure is used to establish situational
awareness systems.

Fig. 1. SOA
munotes.in

Page 18

Cloud Computing –II
18 Grid computing
Grid computing is also known as distributed computing. It is a type of
processor architecture that merges various different computing resources
from multiple locations to achieve a common goal.
Types of machines used in computing:
1. Control Node: It is a group of servers which administers the whole
network.
2. Provider: It is a computer which contributes its resources in the
network resource pool.
3. User: It uses the resources on the network.

Fig.2 Grid Computing
Utility Computing
Utility computing is the bulk trending IT service model. It supply on -
demand computing resources (i. e. computation, storage, and programming
services via API) and infrastructure based on the pay per use method.

Fig.3. Utility Computing munotes.in

Page 19

Computing Platforms -I
19 ● Cloud computing gives guarantee to transform computing into a utility
delivered over the internet.
● Enterprise archit ecture is a function within IT departments that has
developed over time, playing a high value role in managing transitions
to new technologies, such as cloud computing.
2.2 CLOUD COMPUTING DEFINITION AND
CHARACTERISTICS
● Cloud computing refers to diff erent technologies, services, and
concepts.
● It is associated with virtualized infrastructure or hardware on demand,
utility computing, IT outsourcing, platform and software as a service,
and many other things that now are the heart of the IT industry.
● Figure 4 shows too many different ideas included in current definitions
of cloud computing systems.
● The term cloud has an abstraction of the network in system diagrams.
This meaning is also put into cloud computing, which refers to an
Internet -centric way of computing.
● The Internet plays a fundamental role in cloud computing, it shows the
medium or the platform through which many cloud computing
services are delivered and made accessible.
Definition of cloud computing:
“Cloud computing introduces both the a pplications delivered as services
over the Internet and the hardware and system software in the data centers
that provide those services.”
● Cloud computing as aoccurance touching on the entire stack: from the
underlying hardware to the high -level software s ervices and its
applications.
● Here is the concept of everything as a service, like XaaS, the different
components of a system IT infrastructure, development platforms,
databases, and so on can be delivered, measured, and consequently
cost as a service.
● Cloud computing is a model for defining ubiquitous, convenient, on -
demand network access to a shared pool of configurable computing
resources (e.g., like networks, servers, storage, applications, and
services) that can be instantly provisioned and free with less
management effort or service provider interaction.
munotes.in

Page 20

Cloud Computing –II
20

Fig. 4 Cloud computing environment
● Utility -oriented approach is an important aspect of cloud computing
● Cloud computing concentrates on delivering services with a given
pricing model, in most case s a pay -per-use method.
● It is also possible to access online storage, rent the virtual hardware, or
use development platforms and pay only for their effective usage, with
no or minimal up -front costs.
● All these operations can be executed and billed simpl y by entering the
credit card details and accessing the exposed services through a Web
browser.
● The criteria to disfavor whether a service is delivered in the cloud
computing style:
○ The service is accessible through a Web browser or a Web services
applica tion programming interface.
○ Zero capital expenditure
○ pay only for what you use
● Characteristics
○ Resources Pooling
Cloud providers pulled the computing resources to give services to
multiple customers with the help of a multi -tenant model.
○ On-Demand Self -Service
It is one of the key and valuable features of Cloud Computing as the user
can regularly monitor the server uptime, capabilities, and allotted network
storage.

munotes.in

Page 21

Computing Platforms -I
21 ○ Easy Maintenance
The servers are easy to maintain and the downtime is required very less
and even in some situations, there is no downtime.
○ Large Network Access
The user can access the data of the cloud or upload the data to the cloud
from anywhere just with the help of a device and an internet connection.
○ Availability
The potential of the Cloud can be altered as per the use and can be
extended a lot. It studies storage usage and allows the user to purchase
extra Cloud storage if needed for a very small amount.
○ Automatic System
Cloud computing automatically analyzes the data needed and suppo rts a
metering capability at some level of services.
○ Economical
It is a one -time investment as the company has to buy the storage and a
small part of it can be provided to the many companies which save the
host from monthly or yearly costs.
○ Security
Secur ity creates a snapshot of the data stored so that the data may not get
lost even if one of the single servers gets damaged.
○ Pay as you used
In cloud computing, the user has to pay only for the service they have
utilized and used.
○ Measured Service
Cloud Co mputing resources used to handle and the company uses it for
recording.
2.3 ENTERPRISE COMPUTING,
● Enterprise computing defines the use of computers for data processing
in large organizations, also called as ‘information systems’ (IS), or
even ‘information technology’ (IT).
● The key elements of cloud computing, are:
○ Computing resources wrap up as a commodity and made available
over the internet.
○ Ability for end -users to rapidly give the resources they need.
○ a costing that charges consumers only for those cl oud resources they
actually use. munotes.in

Page 22

Cloud Computing –II
22 MAINFRAME ARCHITECTURE
● The mainframe architecture is shown in Figure 5.
● A terminal -based user interface display screens controlled by the
mainframe server using the “Virtual Telecommunications Access
Method (VTAM)”

Fig. 5. Mainframe architecture
● for entering and viewing information. Terminals communicated with
the mainframe using the ‘systems network architecture’ (SNA)
protocol, instead of the ubiquitous TCP/IP protocol of today.
● These mainframe computers had limited CP U power by modern
standards, their I/O bandwidth was extremely generous relative to their
CPU power.
● In Outcome, mainframe applications were constructed using a batch
architecture to minimize utilization of the CPU during data entry or
retrieval.
● In the o ld time mainframe architectures, application data was stored
either in structured files, or in database systems based on the
hierarchical or networked data model.
● The storage subsystem in mainframes, called ‘virtual storage access
mechanism’ (VSAM), built in support for a variety of file access and
indexing mechanisms as well as sharing of data between concurrent
users using record level locking mechanisms.
munotes.in

Page 23

Computing Platforms -I
23 CLIENT -SERVER ARCHITECTURE
● The Client -Server Architecture system shown in Figure 6.
● First, the ‘for ms’ architecture for minicomputer -based data processing
became popular.
● This architecture took a part in the use of terminals to access server -
side logic in C, mirroring the mainframe architecture, PC -based forms
applications provided graphical ‘GUIs’ as opposed to the terminal -
based character -oriented ‘CUIs.’
● The GUI forms model represents the first client server architecture.
● The ‘forms’ architecture evolved into the more general client -server
architecture, wherein significant processing logic executes in a client
application, such as a desktop personal computer.

Fig.6. Client -server architectures
● The client -server architecture is also referred to as a ‘fat -client’
architecture, as shown in Figure 6.
● The client application makes calls using SQL comma nd, to the
relational database using networking protocols., running over a local
area (or even wide area) network using TCP/IP. Business logic largely
stays within the client application code, though some business ideas
can also be implemented within the d atabase for faster performance,
using ‘stored procedures.’
● client -server applications normally made many (client) requests to the
server during the processing of a single screen. This type of request
was large as compared to the terminal, where only the in put and final
result of a computation were transferred.
munotes.in

Page 24

Cloud Computing –II
24 3-TIER ARCHITECTURES WITH TP MONITORS
● Transaction -processing monitors were redeveloped to solve this
problem for midrange database servers.
● These TP monitors were the first examples of ‘middleware,’ which
reside between clients and a database server to manage access to
scarce server resources, essentially by queuing client requests. Thus, as
depicted in Figure 7, by limiting concurrent requests to a small
number, say 50, the server could handle the la rge load while the clients
only paid a small price in response time while their requests waited in
the TP monitor queues.

Fig.7 3-tier architecture scales
● In a TP monitor architecture, the requests being queued were
‘services’ implementing business logic and database operations.
● These were executed as a number of Unix processes, each one is
publishing many such services, typically as remote procedure calls.
● The TP monitor model is also called the 3 -tier architectural model,
where client, business and da ta layers are clearly separated and often
also stay on separate machines, as shown in diagram 8.
munotes.in

Page 25

Computing Platforms -I
25

Fig.8 3 -tier TP monitor architecture
● The 3 -tier model based on ‘object -based’ access to services, replacing
flat remote procedure calls, with the introduct ion of object -oriented
distributed communication systems such as CORBA.
● In CORBA (Common Object Request Broker Architecture), the client
application can communicate with services on the server via methods
on distributed objects in lieu of having to build in application specific
message handling for passing parameters to services and receiving
their responses accordingly.
2.4 THE INTERNET AS A PLATFORM
● The internet is used as a communication infrastructure for data sharing
between large government research labs, and grew to include academic
institutions across the world.
● The internet uses a platform for sharing data, documents, using the
HTTP protocol and HTML.
● Using a browser, information published over the internet could be
accessed unspecified by the pu blic at large, giving rise to the world
wide web (WWW).
INTERNET TECHNOLOGY AND WEB -ENABLED
APPLICATIONS
● Internet -based applications depend on HTTP and HTML both are now
standards defined by the world wide web consortium (W3C). munotes.in

Page 26

Cloud Computing –II
26

Fig.9 . Internet technolog y and web -enabled applications
● As shown in Figure 9, a web server is a process, such as the Apache
HTTPD daemon, that receives HTTP requests from clients, typically
web browsers.
● Requests are lined up until assigned to a request handler thread within
the w eb server process.
● The server sends an HTTP response which contains data, either
recalled directly from a file system or as computed by a server
program initiated to respond to the client request.
● The CGI (common gateway interface) protocol is used by th e web
server to launch server programs and communicate with them, i.e. pass
parameters and accept their results, such as data retrieved from a
database. The browser client purely interprets HTML returned by the
server and displays it to the user.
WEB APPLI CATION SERVERS
● In this application architecture, processing logic, including database
access, took place outside the web server process via the programs
invoked by it.
● To serve HTTP requests from files or CGI scripts, requests could also
be processed from multithreaded execution environments, called
containers carried within the web server.
● The servlet container, first introduced in the pure Java Apache Tomcat
server, which allowed Java programs to execute in a multi -threaded
manner within the server proce ss as Servlet code. munotes.in

Page 27

Computing Platforms -I
27 ● The container also handles load balancing across coming requests
using these threads, as well as database connection pooling, in a mode
similar to TP monitors.
● The application -server architecture is the same advantages of a 3 -tier
architecture, i.e, the ability to operate larger workloads as compared to
the client -server model.

Fig.10. Web application server
● Servlet code used to respond to HTTP type requests.
● JSPs also introduced in Tomcat server, which allowed the user
interface be havior to be encoded directly as Java code embedded
within HTML code. Such ‘JSP’ files are dynamically compiled into
servlets code, as shown on the right in Figure 10.

Fig.11. Web application server technology stacks
munotes.in

Page 28

Cloud Computing –II
28 ● The Microsoft web or application ser ver, IIS (Internet Information
Server), runs only on the Windows operating system.
● The J2EE stack, multiple language support provided, including C,
C++, and Microsoft specific languages such as C# and VB.
● The application container at this point was simpl y Microsoft’s COM
environment on the Windows operating system that allowed multiple
processes to execute and communicate with each other. Recent
versions of this stack are called the .NET framework.
INTERNET OF SERVICES
● The web services and architectures for integration over the internet
● IOS provides a rich client -side interface such as a Google Map, can be
implemented in Javascript that accesses the server over HTTP using
asynchronous requests via XMLHTTPRequest.
● In conventional server -to-server web ser vices, application server code
accesses published services from another server via SOAP over HTTP.

Fig.12 Internet of services
● The software as a service and cloud computing paradigms bring in this
contractual aspect formally, while also re -spotlight the human element.
2.5 SUMMARY
● Virtualization is the process of creating a virtual environment to run
multiple applications and operating systems on the same server.
● Cloud computing refers to both the applications delivered as services
over the Internet and th e hardware and system software in the data
centers that provide those services. munotes.in

Page 29

Computing Platforms -I
29 ● The internet is used as a communication infrastructure for data sharing
between large government research labs, and grew to include academic
institutions across the world.
● In a web -enabled application architecture, processing logic, including
database access, took place outside the web server process via scripts
or programs invoked by it, using CGI for interprocess communication.
● An Internet of Services (IoS) continues, it is o f utmost importance for
the telecoms industry to understand what the IoS is and upon what
foundations and methodologies the IoS is based and built on.
2.6 REFERENCE FOR FURTHER READING
● Enterprise Cloud Computing Technology, Architecture, Applications,
GautamShroff, Cambridge University Press, 2010
● Mastering In Cloud Computing, RajkumarBuyya, Christian Vecchiola
And ThamariSelvi S, Tata Mcgraw -Hill Education, 2013
● Cloud Computing: A Practical Approach, Anthony T Velte, Tata
Mcgraw Hill, 2009
2.7 UNIT END EX ERCISES
1. What is cloud computing? Explain the characteristics of cloud
computing.
2. Write a short note on:
a. Client -Server Architecture.
b. Mainframe Architecture.
3. Explain the internet as a platform in cloud computing.

munotes.in

Page 30

30 3
COMPUTING PLATFORMS -II
Unit Structure :
3.0 Objective
3.1 Introduction
3.2 Cloud computing services: SaaS, PaaS, IaaS,
3.3 Enterprise architecture,
3.4 Types of clouds.
3.5 Summary
3.6 Reference for further reading
3.7 Unit End Exercises
3.0 OBJECTIVE
● To understand the different types of cloud computing servcies.
● To understand the enterprise architecture used in cloud computing.
● To study the different types of clouds.
3.1 INTRODUCTION
● Cloud Computing can be defined as the exercise of using a net work of
remote servers hosted on the Internet to store, manage, and process
data, alternatively a local server or a may be a personal computer.
● Organizations offering such types of cloud computing services are
called cloud providers and charge for cloud c omputing services based
on their usage.
● Grids and clusters are the base for cloud computing.
3.2 CLOUD COMPUTING SERVICES: SAAS, PAAS,
IAAS, SOFTWARE AS A SERVICE (SAAS)
● Software -as-a-Service (SaaS) is a software delivery model that give
access to applica tions through the Internet as a Webbased service.
● It provides a free users from complex hardware and software
management by offloading such tasks to third parties, which build
applications accessible to multiple users through a Web browser. munotes.in

Page 31

Computing Platforms -II
31 ● Example:custo mers neither need to install anything on their premises
nor have to pay costs to buy the software and the need licenses. They
directly access the application website, enter their username and
password and other billing details, and can immediately use the
application.
● On the source side, they keep maintaining specific details and features
of each customer's infrastructure and make it available on user request.
● SaaS as a one -to-many software delivery model, whereby an
application is shared across multiple u sers.
● Example includes CRM3 and ERP4 applications that add up common
needs for almost all enterprises, from small to medium -sized and large
businesses.
● This structure relives the development of software platforms that
provide a general set of features an d support specialization and ease of
integration of new components.
● SaaS applications are naturally multitenant.
● The acronym SaaS was then invented in 2001 by the Software
Information & Industry Association (SIIA).
● The analysis done by SIIA was mainly ali gned to cover application
service providers (ASPs) and all their variations, which imprison the
concept of software applications consumed as a service in a wide
sense.
● ASPs Core characteristics of SaaS:
○ The product sold to customers is an application appr oach.
○ The application is centrally managed.
○ The service delivered is one -to-many.
○ The service provides is an integrated solution delivered on the
contract, which means provided as promised
Platform as a service
● Platform -as-a-Service (PaaS) which provide s a development and
deployment platform for running applications in the cloud.
● They compose the middleware on top of which applications are built.
● Following figure shows a general overview of the features
characterizing the PaaS. munotes.in

Page 32

Cloud Computing –II
32

Fig.1. The Platform -as-a-Service reference model
● Application management is the key functionality of the middleware
systems.
● PaaS implementations provide applications with a runtime
environment and do not expose any service for managing the
underlying infrastructure.
● They auto mating the process of deploying applications to the
infrastructure, configuring application components, provisioning and
configuring supporting technologies such as load balancers and
databases, and managing system change based on policies set by the
user.
● The core middleware is responsible for managing the resources and
scaling applications on demand, according to the adherence made with
users.
● The core middleware exposes interfaces that enable programming and
installing applications on the cloud.
● The Pa aS model provides a complete object model for representing an
application and provides a programming language -based approach.
● In this point the traditional development environments can be used to
design and develop applications, which are then deployed on the cloud
by using the APIs revealed by the PaaS provider.
● PaaS offers middleware for developing applications together with the
infrastructure.


munotes.in

Page 33

Computing Platforms -II
33 Infrastructure as a service or hardware as a service
● Infrastructure as a Service is the most popular model a nd developed
market segment of cloud computing.
● They deliver customizable infrastructure on request.
● The IaaS offering umbrella ranges from single servers to entire
infrastructures, including network devices, load balancers, and
database and Web servers.
● The main aim of this technology used to deliver and implement these
solutions is hardware virtualization:
○ one or more virtual machines configured and interconnected
○ Virtual machines also constitute the atomic components that are
installed and charged ac cording to the specific features of the virtual
hardware:
■ memory
■ number of processors, and
■ disk storage
● IaaS shows all the benefits of hardware virtualization:
○ workload partitioning
■ application isolation
■ sandboxing, and
■ hardware tuning
● HaaS allows better utilization of the IT infrastructure and provides a
more safe environment for executing third party applications.
● Figure 2 shows a total view of the components setup an Infrastructure -
as-a-Service.
● It is possible to identified three principal laye rs:
○ the physical infrastructure
○ the software management infrastructure
○ the user interface
● At the top layer the user interface allow to access the services exposed
by the software management infrastructure. This types of an interface
is generally based on Web technologies:
munotes.in

Page 34

Cloud Computing –II
34 ○ Web services
○ RESTful APIs, and
○ mash -ups
● These automation allow applications or final users to access the
services exposed by the underlying infrastructure.
● A central role of the scheduler, is in charge of allocating the execution
of virtual machine instances. The scheduler communicates with the
other components that perform a variety of tasks.

Fig.2. Infrastructure -as-a-Service reference implementation
3.3 ENTERPRISE ARCHITECTURE
● The ‘enterprise architecture’ function within enterpri se IT has evolved
to manage the complexities of an ever -changing technical
environment.
● In the process enterprise architects found it useful to maintain a
description of all of an enterprise’s software applications, how they
fulfill business needs, how th ey are implemented technically and how
they communicate with each other.
ENTERPRISE DATA AND PROCESSES
● The first step is identifying and naming each process, and identifying
business events that mark its start and end.
● Enterprise processes can often be cl assified as ‘vertical’ or
‘horizontal’. munotes.in

Page 35

Computing Platforms -II
35 ○ Vertical processes typically operate within a single organizational
function, such as sales or accounting, manage a cohesive information
set pertaining to that function, and are typically supported by software
packa ges or systems dedicated to that department. ‘Prospect to order’,
for example, is a vertical process limited to the sales function.
○ Horizontal processes, on the other hand, cut across functional units;
‘order to cash’ is a horizontal process since it span s sales, production
and finance.
ENTERPRISE COMPONENTS
● The term ‘component’ has traditionally been taken to mean a ‘software
sub-system that has well -defined interfaces which can be used
independently of its internal implementation’ [50]. Structuring
softw are into components drives modularity in software development
and makes it easier to evolve a large system by incrementally
replacing its components over time.
Figure 4.1 shows this component view of enterprise architecture.

Fig 3. Enterprise components
APPLICATION INTEGRATION AND SOA
● Application integration is the often perceived need for a unified view
of data residing in disparate application systems, say for end -to-end
process monitoring, real -time decision support, or for data
warehousing and busine ss intelligence.
There are a number of mechanisms that applications can use to
communicate with each other at different ‘levels’ munotes.in

Page 36

Cloud Computing –II
36 1. Data level integration: direct data transfer using batch programs or
on-line exchange using database triggers
2. API level integra tion: applications publish API libraries that are used
by other applications to access their data
3. Service -method -level integration: applications publish services using
say, web service protocols, in an organized manner so that many
different applications can use a particular service
4. User interface level integration: applications publish mashup APIs
that are used to provide a common user interface to functionality from
many applications
5. Workflow level integration: tasks performed in one application lead
to work items being created in others, thereby driving the flow of work
in a business process
ENTERPRISE TECHNICAL ARCHITECTURE
The definition and management of standards defining the technical
architecture, tools and technical components used in an enterpr ise:
1. Unformity
● Cost and simplicity is the motivation for standardizing technical
components such as application servers, databases and integration
tools used in an enterprise.
● This ‘uniformity’ approach is clearly sound when most of the
integration betwee n disparate technologies is carried out by the
enterprise itself.
2. Network and data security
● Security issues arising from technology choices are also part of the
enterprise architecture function. While considering the option of cloud
deployment the questio n of security of data that will reside outside the
enterprise data center is a common concern
● Network security, or rather securing applications on the network, is a
more serious concern when considering cloud deployment.
3. Implementation architectures and q uick-wins
● One of the aspects enterprise architects pay attention to are the
‘implementation architectures’ required for adopting any new
technology.
● These include the people skills required, along with development,
testing and deployment tools needed, as well as the impact on
business -continuity and disaster -recovery environments.
munotes.in

Page 37

Computing Platforms -II
37 3.4 TYPES OF CLOUDS
● Clouds compose the primary outcome of cloud computing.
● They are a type of parallel and distributed system, physical and virtual
computers conferred as a u nified computing asset.
● Clouds build the infrastructure on top of which services are
implemented and delivered to customers. Such infrastructures can be
of different types and provide useful information about the nature and
the services offered by the clo ud.
● A more convenient classification is given according to the
administrative domain of a cloud: It identifies the boundaries within
which cloud computing services are implemented, provides hints on
the underlying infrastructure take on to support such se rvices, and
qualifies them. It is then possible to evolve four different types of
cloud:
● Public clouds.
○ The cloud is open to the wider public.
● Private clouds.
○ The cloud is executed within the private property of an institution and
generally made accessib le to the members of the institution
● Hybrid clouds.
○ The cloud is a combination of the two previous clouds and most likely
identifies a private cloud that has been augmented with services hosted
in a public cloud.
● Community clouds.
○ The cloud is disting uished by a multi administrative domain consisting
of different deployment models (public, private, and hybrid).
Public clouds
● Public clouds account for the first expression of cloud computing.
● They are an awareness of the canonical view of cloud computin g in
which the services provided are made available to anyone, from
anywhere, and at any time through the Network.
● From a structural point of view they are a distributed system, most
likely composed of one or more data centers connected together, on
top o f which the specific services offered by the cloud are
implemented.
● Any customer can easily agree with the cloud provider, enter her
username and password and billing details. munotes.in

Page 38

Cloud Computing –II
38 ● They offer results for minimizing IT infrastructure costs and serve as a
viable option for handling peak loads on the local infrastructure.
● They are used for small enterprises, which are able to initiate their
businesses without large up -front investments by completely relying
on public infrastructure for their IT needs.
● A public c loud can recommend any type of service such as
infrastructure, platform, or applications. For example, Amazon EC2 is
a public cloud that provides infrastructure as a service; Google
AppEngine is a public cloud that provides an application development
platform as a service; and SalesForceservice.com is a public cloud that
provides software as a service.
Private clouds
● private clouds, which are the same as public clouds, but their resource
provisioning model is restricted within the boundaries of an
organizat ion.
● Private clouds have the benefit of keeping the core business operations
in house by depending on the existing IT infrastructure and reducing
the cost of maintaining it once the cloud has been set up.
● The private cloud can provide services to a differe nt range of users.
● private clouds is the possibility of testing applications and systems at a
comparatively less price rather than public clouds before implementing
them on the public virtual infrastructure.
● The main advantages of a private cloud computi ng infrastructure:
1. Customer information protection.
2. Infrastructure ensuring SLAs.
3. Compliance with standard procedures and operations.

Fig.4. Private clouds hardware and software stack. munotes.in

Page 39

Computing Platforms -II
39 Hybrid clouds
● A hybrid cloud could be an attractive opportunity fo r taking advantage
of the best of the private and public clouds. This shows the
development and diffusion of hybrid clouds.
● Hybrid clouds enable enterprises to utilize existing IT infrastructures,
maintain sensitive information within the area, and natural ly increase
and reduce by provisioning external resources and releasing them
when they’re no longer needed.
● Figure 5 provides a general overview of a hybrid cloud:
○ It is a heterogeneous distributed system resulting from a private cloud
that integrates ad ditional services or resources from one or more public
clouds.
○ For this reason they are also called heterogeneous clouds.
○ Hybrid clouds address scalability issues by leveraging external
resources for exceeding

Fig.5. Hybrid/heterogeneous cloud overview.
Community clouds
● Community clouds are distributed systems created by integration of
services of different clouds to handle the specific requirement of an
industry, a community, or a business sector.
● The National Institute of Standards and Technologies (NI ST)
characterizes community clouds as follows:
○ The infrastructure is shared by different organizations and supports a
certain community that has shared concerns. munotes.in

Page 40

Cloud Computing –II
40 ○ It may be controlled by the organizations or a third party and may exist
on premise or off pr emise.

Fig. 6 general view community clouds
● Figure 6 shows a view of the usage scenario of community clouds,
jointly with reference architecture.
● The users of a distinct community cloud fall into a well identified
community, sharing the same concerns or needs such as government
bodies, industries, or even simple users, but all of them concentrate on
the same problem for their interaction with the cloud.
● Community clouds are the services that are generally delivered within
the institution that owns the c loud.
● Candidate district for community clouds are as follows:
○ Media industry
■ Where Companies are finding low -cost, agile, and simple solutions to
better the efficiency of content production.
■ Most media involve an expanded ecosystem of partners.
■ Communit y clouds can provide a shared environment where services
can ease business to business participation and give the horsepower in
terms of aggregate bandwidth, CPU, and storage required to efficiently
support media production.
○ Healthcare industry.
■ In the h ealthcare industry, there are different storyline in which
community clouds are used. munotes.in

Page 41

Computing Platforms -II
41 ■ Community clouds provide a global platform on which to share
information and knowledge without telling sensitive data maintained
within the private infrastructure.
■ The naturally hybrid deployment model of community clouds supports
the storing of patient data in a private cloud while using the shared
infrastructure for noncritical services and automating processes within
hospitals.
○ Energy and other core industries.
■ Thes e industries concern different service providers, vendors, and
organizations, a community cloud can give the right type of
infrastructure to create an open and upright market.
○ Public sector.
■ The public sector can limit the adoption of public cloud offeri ngs.
■ governmental processes involve several institutions and agencies
■ Aimed at providing strategic solutions at local, national, and
international administrative levels.
■ involve business -to-administration, citizen -to-administration, and
possibly business -to-business processes.
■ Examples, invoice approval, infrastructure planning, and public
hearings.
○ Scientific research.
■ THis is an interesting example of community clouds.
■ In this point, the common interest in handling and using different
organizations t o split a large distributed infrastructure is scientific
computing.
The Advantages of community clouds:
● Openness.
Clouds are open systems in which fair competition between different
solutions can occur.
● Community.
Providing resources and services, the i nfrastructure turns out to be more
scalable.
● Graceful failures.
There is no single provider & vendor in control of the infrastructure, there
is no chance of a single point of failure.
munotes.in

Page 42

Cloud Computing –II
42 ● Convenience and control.
There is no dispute between convenience and control because the cloud is
shared and owned by the community, which makes all the decisions
through a collective representative process.
● Environmental sustainability.
These clouds tend to be more organic by increasing and shrinking in a
symbiotic relat ionship to support the demand of the community.
3.5 SUMMARY
● Three service models. Software -as-a-Service (SaaS), Platform -as-a-
Service (PaaS), and Infrastructure -as-a-Service (IaaS).
● Four deployment models. Public clouds, private clouds, community
clouds, a nd hybrid clouds.
● Cloud computing has been rapidly adopted in industry, there are
several open research challenges in areas such as management of cloud
computing systems, their security, and social and organizational
issues.
3.6 REFERENCE FOR FURTHER READI NG
● Enterprise Cloud Computing Technology, Architecture, Applications,
GautamShroff, Cambridge University Press, 2010
● Mastering In Cloud Computing, RajkumarBuyya, Christian Vecchiola
And ThamariSelvi S, Tata Mcgraw -Hill Education, 2013
● Cloud Computing: A Pr actical Approach, Anthony T Velte, Tata
Mcgraw Hill, 2009
3.7 UNIT END EXERCISES
1. What does Infrastructure -as-a-Service refer to?
2. What are the main characteristics of a Platform -as-a-Service solution?
3. What does the acronym SaaS mean? How does it relate to c loud
computing?
4. Classify the various types of clouds.
5. Give an example of the public cloud.

munotes.in

Page 43

43 4
CLOUD TECHNOLOGIES -I
Unit Structure
4.0 Objectives
4.1 Introduction
4.2 Cloud computing platforms
4.2.1 Infrastructure as a service: Amazon EC2
4.2.2 Platform as a service: Google App Engine
4.2.2.1 Google Datastore
4.2.2.2 Amazon SimpleDB
4.2.3 Mic rosoft Azure
4.3 Web services
4.3.1 Web Services: SOAP and REST
4.3.1.1 SOAP/WSDL Web services
4.3.1.2 REST web services
4.3.2 SOAP VERSUS REST
4.4 AJAX
4.5 Mashups
4.6 Multi -tenant software
4.6.1 Multi -entity support
4.6.2 Multi -schema approach
4.6.3 Multi -tenancy using cloud data stores
4.6.4 Data access control for enterprise applications
4.7 Summary
4.8 List of References
4.9 Unit End Exercises

munotes.in

Page 44

Cloud Computing –II
44 4.0 OBJECTIVES
The objectives of this chapter are:
 To understand the concept of cloud computing
 To ap preciate the evolution of cloud from the existing technologies
 To have knowledge on the various issues in cloud computing
 To be familiar with the lead players in cloud
 To appreciate the emergence of cloud as the next generation
computing paradigm
4.1 INTRO DUCTION
The provision of computing resources as a service is known as cloud
computing. Going to the cloud essentially means that a provider rather
than the end user owns and manages the resources.Through their internet -
connected PCs, smartphones, tablets, and wearables, users of cloud
computing technologies can access storage, files, applications, and servers.
Data is processed and stored by cloud computing services away from end
users.
In its simplest form, cloud computing refers to the capability of stori ng
and accessing data and software through the internet as opposed to a hard
drive. This implies that organisations of any size may use robust software
and IT infrastructure to grow, get leaner, and become more flexible while
competing with much larger fir ms. Contrary to traditional hardware and
software, cloud computing enables companies to stay on the cutting edge
of innovation without having to make significant financial commitments
in their own equipment purchases, upkeep, and repairs.
4.2 CLOUD COMPUTI NG PLATFORMS
We will discuss the main platforms from Amazon, Google, and Microsoft
in this chapter, summarising the services they offer from the viewpoint of
end users.We will also discuss more cloud service providers software&
resources used for set up.
4.2.1 Infrastructure as a service: Amazon EC2
Through an automated web -based administration panel, they offer IaaS,
which allows users to virtually provision computing infrastructure within
fraction of time. Elastic Compute Cloud (EC2) is one of several Iaa S
services that make up the platform.
Facilities offered are depicted in Figure 1. The picture depicts a very vast
network of servers that are used to implement these services as dashed
boxes. Users of the Elastic Compute Cloud service can use dedicated
virtual machines with the end -user being unaware of the physical server's munotes.in

Page 45

Cloud Technologies -I
45 specifics, like as its location, capacity, etc. Users create PKI1 key pairs
through the administration dashboard, which they can use to securely
access these virtual servers online.

Figure 1: Amazon infrastructure cloud
In Figure 1, person C configures VM4 using the management interface.
For a Linux server, ssh is used to access it. For a Windows server, remote
desktop is used. When provisioning a server, users can select from a
varie ty of VM (Amazon machine images, or AMIs). All AMIs are used to
boot the required configuration of virtual servers.The user's account is
charged hourly depending on actual use, or the amount of time the server
is operational. The AMI that is utilised and t he server capacity that is
selected while provisioning affects the costs.Cloud users have full control
over these servers because they have root or administrator access to them.
Numerous servers that can connect with one another across the quick
internal n etwork of the Amazon cloud can be provisioned and accessed by
users. In Figure 1, for instance, user C has also provisioned VMs 5 and 6
in addition to VM4. VM5 may be a database server if VM4 is a web
server, and the two of them can communicate using TCP/I P across the
internal cloud network. As a database server, VM5 must be able to save
and fetch information. With the help of the SimpleDBfacility, key -value
pairs can be quickly saved and retrieved in an object store. The distinction
between SimpleDB and re lational databases must be made.
Businesses looking to use cloud computing must also solve corporate IT
security issues. One of these issues, network security, is crucial: Firewalls,
proxies, intrusion detection systems, and other security measures are
typically used to protect an organization's computing resources.
Obviously, safety demands that VM operating in the cloud be safeguarded,
employing similar guidelines & security measures. With the help of a
VPN, servers could be linked to an organization's ne twork utilising the
cloud service offered by Amazon EC2 (virtual private network).Users A
and B, for instance, connect to virtual servers VM1, VM2, and VM3 via a munotes.in

Page 46

Cloud Computing –II
46 VPN that is active through online platform in Figure 1. The network
operations centre for the company then manages the private IP addresses
assigned to these machines.
4.2.2 Platform as a service: Google App Engine
Platform as a Service (PaaS), sometimes known as Google App Engine, is
a service provided by Google. A PaaS solution conceals the real execution
environment from customers. Instead software platform and an SDK are
offered, allowing customers to create and upload apps to the cloud. The
PaaS platform is in charge of operating the applications, handling external
service requests and schedul ing jobs that are part of the application.A
PaaS platform can distribute application servers among users that want
lower capacity and automatically scale resources allotted to apps that
encounter high loads by hiding the real execution servers from the use r.
The Google App Engine user interface is shown in Figure 2. Users upload
files relevant to their codein either Java or Python.

Figure 2: Google App Engine
A PaaS is utilised and made available across the worldthroughout a day,
but it is only charged wh en accessed (or if batch jobs run). Additionally,
the development and testing of applications in Google App Engine is free,
subject to use restrictions; charges only apply when a sufficient number of
requests actually visit facility.Installed programmes th at aren't used just
require storage for their code and data and don't take any CPU time.
In Google's data centres, a sizable number of web servers that process
requests from end users all around the world serve GAE apps. The web
servers process these reque sts by loading GFS code into memory.There is
no assurance that the same server will fulfil more than one request, even if
they originate from the same HTTP session. Every access to a specific munotes.in

Page 47

Cloud Technologies -I
47 application is handled by any servers.Applications can also indi cate which
operations should be performed in batches by a scheduler.
This architecture makes it possible for programmes to grow naturally as
load rises, but it also makes it difficult for application code to rely on in -
memory data. To partially address thi s issue, Memcacheis provided: It is
specifically used to implement HTTP sessions so that requests from the
same session can often retrieve their session data even if they are sent to
separate servers.
4.2.2.1 Google Datastore
It is a database where applica tions data are kept.They define types
(referred to as "kinds") using the Datastore, and their instances (referred to
as "entities") can be distributedly stored on the GFS file system. There are
significant differences as shown in Fig 3.

Figure 3: Google datastore
All entities of a "kind" need not have the same properties. Instead, every
entity can have extra characteristics added to it. This function is very
helpful when it is impossible to predict all of the possible attributes. A
model for storing 'pro ducts' of various entities must permit
everycategoryfor unique capacity of attributes. This would most likely be
implemented in a relational model utilising a distinct FEATURES table,
reflected on the lower left of Fig 3.This table ('type') is not necessar y when
using the Datastore; rather, each product object may be given a unique set
of characteristics during runtime.
The Datastore supports basic conditional queries, like the one in Figure
4.3 that returns all customers with names that fall within a certa in lexical
range. With a few limitations, the query syntax is substantially similar as
SQL. A query that included screened clients based on, prohibited in GQL
& permitted in SQL since all unequal conditions in a query must be on a
single property. munotes.in

Page 48

Cloud Computing –II
48 Here GFS distributed file system and several servers are used to keep the
objects (entities) of all GAE apps. From the user's perspective, it's crucial
to make sure that information is (a) properly fetched & (b) adapted despite
being shared with numerous other use rs in a distributed storage
system.The Datastore offers a method for classifying entities into
hierarchical groups that can be utilised for both of these objectives.
4.2.2.2 Amazon SimpleDB
SimpleDB is similarly a database where "domains" are equivalent to
"kinds," and "items" are equivalent to entities. SimpleDB domain queries
support conditions on any number of attributes, including inequality
criteria. Additionally, joins are not allowed, much like in the Google
Datastore. However, unlike Google Datastor e, SimpleDB does not support
object associations.
It's crucial to remember that, like GFS, all information in this case is
copied for layoff. Informationpropagate to minimum one replica, thanks to
SimpleDB's "eventual consistency" concept. This can give th e appearance
of consistency as it's not always the case that reading something right after
after writing it will produce the same outcome. In contrast, writes in
Google Datastore only succeed once all replicas have been updated; this
prevents consistency b ut slows down writing.
4.2.3 Microsoft Azure
Azureservice lately made available to the public on a commercial basis,
while a community preview beta has been accessible to the public.
Azure is a PaaS service, much like Google App Engine. Microsoft
developm ent tools (such as Visual Studio and the supported languages,
C#, Visual Basic, ASPs, etc.) are used by developers to create
applications; very similar to how they are developed and deployed using
Google App Engine's SDK.Similar to Amazon S3, Azure also su pports the
storage of unrestricted files and objects. It also offers a queuing service
like Amazon SQS.
The user's perspective of Microsoft Azure is shown in Figure 4. The
Windows Server operating system is installed on each of the virtual
computers (refer red to as instances) that run the application code when it
is delivered to Azure.Users can indicate the nos of examples a specific
programme will use, but they have no control over when instances boot or
how long they remain active. As a result, compared t o Google App
Engine, Azure offers control level but not the same extent as an IaaS
provider like Amazon EC2. munotes.in

Page 49

Cloud Technologies -I
49

Figure 4: Microsoft Azure
On Azure, program is implemented as a worker or web role.A server that
is built into runs code in role instances. Blobs , queues, and non -relational
tables (similar to SimpleDB) can all be stored in Azure data storage.
Additionally, SQL Data Services in the Azure beta edition included a
different database -storage system. Authorities are Microsoft -owned data
centres where SQ L is deployed on a no of groups. Similar to SimpleDB or
Google Datastore, each container contains entities that have attributes, and
the types and numbers of properties for each item can vary. A SQL Azure
database can only be less than 10 GB in size. Multi ple virtual database
instances must be provisioned in the event that higher data volumes need
to be stored. Furthermore, since joins on larger data sets must be
implemented in application code because cross -database searches are not
allowed.
Microsoft Azur e also offers what are referred to as.NET services in
addition to compute and storage capabilities. These include web -based
workflow orchestration services that may be customised offer across
enabling globally published service end points. They are deploye d on
Microsoft's midware technologies just like SQL Data services and SQL
Azure.
They have benefit of a sizable base apps that are currently in use within
businesses, as opposed to Google's PaaS service.It is expected to get even
simpler in the future to m ove current Microsoft applications to Azure.
Therefore, it's possible that Microsoft Azure overtakes Google's App
Engine as the preferred PaaS platform, at least among large businesses.
4.3 WEB SERVICES
The backbone for web -based cloud computing services i s the set of
protocols that make up the internet: HTTP, DNS, and TCP/IP. This
chapter delves at three crucial web -based technologies that have proved
crucial in enhancing the applicationsadaptability.
munotes.in

Page 50

Cloud Computing –II
50 4.3.1 Web Services: SOAP and REST
In this we demonstr ate outwork of both as well as a comparison in
creating cloud -based offerings.
4.3.1.2 REST web services
The HTML -based web itself is an example of the Representational State
Transfer (REST) architectural style, which was initially developed for use
in lar ge-scale systems with distributed resources.
These are nothing more than HTTP queries to URIs, utilising exact 4
methodologies that the HTTP protocol permits: GET, POST, PUT, and
DELETE. Each URI uniquely detects database record.This service
delivers the c ustomer record in XML format. Links to related recordslook
somewhat like this: http://x.y.com/account/334433, if the record has links
(foreign keys) to related entries.As an alternative, you can directly access
these connections by using data handled in th is "RESTful" fashion in
order to retrieve data.
The above example, as well as two actual instances employing Yahoo! and
Google, who both offer the service are used to illustrate REST web
services in Figure 6. Observe that URLs contain parameters, this fac ility
description vary from paradigm, which stipulates that data should only be
associatedthrough URIs.

Figure 5: REST Web services
4.3.2 SOAP VERSUS REST
Many debates about encoding makes it difficult to provide a web service's
semantics such that it ma y be readily and broadly understood, potentially
allowing multiplefacilitators to provide similar opportunity.The best
illustration is search. REST has the capacity to achieve munotes.in

Page 51

Cloud Technologies -I
51 standardisation:Multiple providers could make this available if e.g.,the
REST st d to searching was http://provider -URL>/query -string>. The reply
doc in XML then be self -descriptive adhering publicly specified top -level
schema while making use of provider -specific name spaces when
necessary. Our examples make it clear right away how mu ch easier it is to
create and use REST -based services than it is to use the more complicated
SOAP/WSDL method.Furthermore, by adopting alternatives like JSON,
REST can avoid costly XML parsing. In light of this, our opinion is that
REST should be the prefe rred option from both a simplicity and an
efficiency standpoint, with the argument for using SOAP/WSDL needing
to be expressly presented based on the circumstances.
Table 1 compares these on six different criteria, including the location of
the servers tha t can provide the service, the level of interaction security,
whether transactions can be supported, how reliant on HTTP technology
the protocol is, the amount expected productivity using each. This
examination leads us to the conclusion that they are easi er, effective& less
expensive to create than SOAP for the majority of requirements. It is clear
that SOAP is being replaced by REST, especially in cloud environments:
The REST API utilising JSON has mostly taken the place of the outdated
Google SOAP servic e.Although SOAP and REST APIs are both published
by Amazon Web Services, SOAP APIs are hardly ever used. In the near
future, REST will also gain tools to address more complicated
functionality, such as transactions.
Table 1: SOAP/WSDL Vs REST
munotes.in

Page 52

Cloud Computing –II
52 4.4 AJAX: A SYNCHRONOUS RICH INTERFACES
Traditional web applications send a series of HTTP GET and POST calls
to their server components, refreshing the HTML page in the browser as
they do so. Client -side JavaScript, or in -browser JavaScript, is only used
for certain cases including animations, hiding or revealing page portions,
and field validations. The browser is largely inactive barring any such
modifications.
Without refreshing their main HTML page, JavaScript programmes that
perform calls using AJAX.A "richer" u ser experience can be achieved by
multiplexing client -side processing of user activities with heavy server -
side processing to reduce the overall response time. Additionally, client -
side Script can issue requests to other internet services in addition to th e
main web server, facilitating the integration of applications within
browsers.
From the standpoint of S/W design, applications are not clients. Instead,
they perform a significant amount of processing on the client, making use
of the desktop's processing capacity exactly like client -server programmes
did. Recall that while employing the client -server architecture, there was a
chance that the interface would be combined in the code, increasing the
difficulty of maintaining such software. Identical issues c ome up when
employing the AJAX paradigm.
In contrast to conventional online applications, AJAX applications —also
known as "rich internet applications" (RIA) —show how they operate in
Figure 7.The remaining portions of the user interface are loaded along
with a base HTML page. The "rich" user interface produced by this
JavaScript software frequently resembles a classic client -server
application. REST web services are used to make asynchronous requests
for data from the server, which the server responds to wit h JSON
structures that the JavaScript code running in the browser can use right
away.

Figure 6: Rich internet application with AJAX munotes.in

Page 53

Cloud Technologies -I
53 AJAX makes it feasible to create user interfaces that are entirely browser -
based and extremely interactive. With the use o f this method, SaaS apps
can start givinginsights resembling programmes, often used within
businesses. This makes SaaS products more appealing to business
customers. Additionally, rather than relying on more complicated
strategies.Furthermore, unlike serve r-side integration, which must be
carried out by corporate IT, simple JavaScript -based integrations are
frequently carried out by business units themselves, in a way similar to
how Dev 2.0 platforms enable end users to create straightforward
applications.
4.5 MASHUPS: USER INTERFACE SERVICES
Mashups raise the bar of integration by incorporating both the service and
the presentation layer for remote services.
Figure 8 uses the Google search service to show mashups. A developer
just needs to acknowledgeand m ake application using codeaccesses
through browser in order to display a Google search bar. This code offers
a "class" called google that has as its methods the Google AJAX API.
After the HTML page has loaded, the methods are calledfor construction.
Observ e that user cannot see an AJAX controller or a REST service; all of
this is concealed within the API methods.Since there doesn’t need
serialisation& thus acceptable in various services.

Figure 7: Mashup example
From the user's point of view, mashups simp lify the process of consuming
web services. The original service request not necessarily be REST; it can
be a customised communication.
However, mashups' requirement to download and run foreign code raises
legitimate security issues, particularly in busin ess settings.In contrast to
ActiveX components, which after installation have virtually full access to
the desktop, JavaScript code often has access to only the browser and the
network, giving the impression that there is no significant security issue.
How ever, in the future, this might no longer be the case: For instance,
Google Gears is a framework that allows apps to run offline by caching
data on the client computer.Although less significant than ActiveX
controls, this poses a potential security risk. munotes.in

Page 54

Cloud Computing –II
54 4.6 MULTI -TENANT SOFTWARE
Applications are typically created for a single organisation, and similarly,
enterprise software is also created so that it may be independently
deployed in each customer's data centre. These programmes often access
and create data that belongs to a single company. Multi -tenancy refers to
the requirement for a solo code to execute on the information of several
clientsin hosted SaaS platforms. This section looks at various approaches
to incorporate multi -tenancy in S/W.
Here, we look towards alternatives to employing virtual machines at the
system level to accomplish multi -tenancy, such as application software
architecture. As a result, this category is sometimes called as application -
level virtualization. Goal of both multi -tenancy a nd virtualization is
resource sharingkeeping users apart of one another.
4.6.1 Multi -entity support
Large globally dispersed enterprises frequently need their systems to
handle various organisational quantities in a sequential and well managed
form long b efore ASPs and SaaS. Assume a bank with numerous branches
that needs to specific S/W to centralised one. It is obvious that the
programme can’t be utilised to access information from multiple sites:
Users for instance, must only view information relevant t o its site.If the
system needed to be improved, for as by adding a new field, the update
would have to be supported by each site at the same instance supporting
modifications. These prerequisites are nearly identical to those for multi -
tenancy!There are ad ditional requirements in a multi -entity scenario, such
as the need to grant a subset of users or a selection of sites based on rank
in an organisational behaviour. The advantages of data centralization
might not be achieved without supporting some global p rocessing.
Figure 9 illustrates modifications that must be done to an application in
order to support fundamental multi -entity features and limit user access to
data. Each record has a field (OU_ID) that identifies the organisational
unit to which each dat a record belongs.A situation mentioning managed
information based on user who is logged in must be added to each
database query. Support for multi -tenancy can be provided using an
identical technique, with the customer to who’s the data records belong
now represented by the OU ID.

Figure 8: Multi -entity implementation munotes.in

Page 55

Cloud Technologies -I
55 The single schema model was widely used in early SaaS deployments,
particularly by those who created their SaaS systems from scratch. The
ability to update application functionality, for as by adding a field, for all
customers at once is one benefit of the single schema structure. However,
there are drawbacks as well: The single schema technique requires
extensive re -structuring of the application code when re -engineering an
established appli cation. This cost may be too high for a large software
programme with millions of lines of code. Additionally, while changes
employing a single schema structure to accommodate customer -specific
extensions like custom fields becomes challenging.It is necess ary to keep
separate databases for the data in these additional fields as well as the
meta -data characterising such adaptations. The application code must
nevertheless read such meta -data in order to, for example, display and
manage custom fields on the sc reen. Fig 10 that shows multipletenant
design model utilising a single schema that providesentries exemplifies
some of these problems more succinctly:

Figure 9: Multi -tenancy using single schema
A Custom Fields table houses metadata and data values for all of the
application's tables. Typically, variations of this method are used to handle
custom fields in a single schema design. Take a look at a screen that is
employed to access and modify Customer table records. The important
record is first fetched us ing name and appropriately selected using the OU
attribute. The custom fields and values are then obtained for specific
entity.For instance, there are two custom fields in OU 503 as they are
shown on the screen, but only one in OU 490 and none elsewhere.
Additionally, certain records might not contain values for these fields.
The aforementioned scenario is an easy one, but there are also more
difficult needs that must be handled, like displaying a list of records
having option to select customisedfields.
4.6.2 Multi -schema approach
It is sometimes simpler to change an existing application to use numerous
schemas, as are supported by the majority of relational databases, rather
than insisting on a single schema. In this paradigm, the programme
determines whi ch OU the user who is now signed in belongs to before munotes.in

Page 56

Cloud Computing –II
56 connecting to the proper database schema. Figure 11 depicts such a
structure.

Figure 1 0: Multi -tenancy using multiple schemas
A different database schema is kept for each client in the multiple schema
approach so that each schema can directly apply customer -specific
customizations. The core schema's meta -data is likewise kept in a separate
database, but unlike the Custom Fields table in Figure 10, this table only
contains meta -data and does not contain field values for specific records.
As a result, the application design is simpler, and it is likely that there will
be fewer and simpler changes required to re -engineer a legacy programme
for multi -tenancy.
Using a multiple schema strategy, think about im plementing the Edit
Customer screen that was previously discussed: The application uses data
from the Meta -Data table to render the appropriate fields on the screen.
Before issuing data manipulation (i.e., SQL) statements during a database
query, the appli cation sets the database schema in order to access the
proper schema. Be aware that implementing components of an
interpretative architecture, which is very similar to the Dev 2.0 paradigm,
is necessary to enable the many schema model.
4.6.3 Multi -tenancy using cloud data stores
Non-relational storage models are present in cloud data repositories.
Additionally, each of these data stores is multitenant from the ground up
because a single instance of a large -scale distributed data store
successfully supports numerous applications developed by cloud
customers. While each Google App Engine user can only develop a certain
number of applications, each of which looks to have its own data store, the
underlying distributed infrastructure is the same for all Google Ap p
Engine users.
Here, we concentrate on a different issue: How does a user (application
developer) build their own multi -tenant application on a cloud
platform?The answer is simple in the case of Amazon EC2 because this
infrastructure cloud provides users with direct access to (virtual) servers
that can be used to replicate the multi -tenant architectures that were munotes.in

Page 57

Cloud Technologies -I
57 previously mentioned utilising common application servers and database
systems.
However, utilising a PaaS platform like Google's App Engine with its
Datastore, Amazon's SimpleDB, or even Azure's data services, the
scenario is different. One data store name space, or schema, for each App
Engine application, for instance, means that if we establish one model
named "Customer," we cannot construct ano ther model with the same
name in the same application. Therefore, it initially seems as though we
are forced to employ the ineffective single schema strategy.
Entities in the Google Datastore, however, are effectively schema -less,
which is an intriguing fe ature. As a result, the language API is in charge of
defining how the data storage is used. Particularly, the dynamic and
object -oriented Model class in the Python API for App Engine. All entities
of a "kind" acquire their properties from a class definitio n that derives
from the Model class. Additionally, because Python is an entirely
interpretative language, new classes and the accompanying data storage
"kinds" can be defined at runtime.
Figure 12 illustrates a hypothetical Python implementation of multi -
tenancy with Google App Engine utilising different schemas. At runtime,
distinct classes are instantiated for each schema. By using table names that
depend on the schema, this method is comparable to simulating several
schemas in a relational database.

Figure 1 1: Multi -tenancy using google datastores
A similar approach may be utilised with Amazon's SimpleDB, where
domains, which are the relational equivalent of "kind" in the Google
Datastore and act as tables in relational terminology, can be generated
dynamically using any of the supplied language APIs.

munotes.in

Page 58

Cloud Computing –II
58 4.6.4 Data access control for enterprise applications
We have discussed the traditional methods for achieving multi -tenancy so
far from the viewpoint of allowing a single application code base to deal
with data from numerous customers while operating in a single instance,
hence reducing administrative expenses across a potentially large number
of customers.
Multi -tenancy as it was previously mentioned seems to be most useful in a
software as a service para digm. Additionally, there are several situations
when multi -tenancy might be advantageous for the company. Supporting
various entities, like bank branches, is fundamentally a multi -tenancy
need, as we have already seen. If a workgroup -level application mus t be
distributed to numerous autonomous teams, who often do not need to
share data, similar needs may arise. In certain situations, modifications to
the application schema may be required to handle differences in business
processes.Supporting various legal entities, each of which may operate in
a distinct regulatory environment, also raises similar requirements.
As we previously indicated, based on their position in the organisational
unit hierarchy, a subset of users in a multi -entity scenario may need acc ess
to data from all branches or a selection of branches. In a broader sense,
access to data may need to be restricted based on the values of any field in
a table, such as making certain users the only ones who can see high -value
transactions or making som e customer names opaque unless given specific
permission.These requirements, known as data access control demands,
are widespread but less frequently handled in a reusable and generic way.
Data access control, often known as DAC, is a generalisation of mul ti-
tenancy because it may frequently be used to implement the latter.Figure
13 shows how data access control can be implemented generically within
a single schema to allow relatively general rules for restricting access to
records depending on field values .

Figure 1 2: Data access control munotes.in

Page 59

Cloud Technologies -I
59 Every application table, including Customer, has a new field called DAC
ID added to it. The values of the DAC ID in each Customer record are
populated through a batch process using patterns based on value ranges of
arbitr ary variables that are listed in the DAC Rules table. According to the
information provided in the User DAC Roles table, users are given the
authority to access records that satisfy one or more of these DAC rules. A
batch method is used to expand this data to the User table, which contains
records for each value of the DAC ID that a user can access.For instance,
the User table contains five records when the User 101 has access to three
DAC rules. This calculation determines which subset of these a specific
user has access based on the User DAC Roles information after computing
the entire set of mutually exclusive and unique DAC range combinations
based on the DAC Rules; take note that this computation is independent of
the actual DAC ID values in the Custome r or other application tables.It is
simple to use a join to restrict access to Customer table data to only those
that a specific user is allowed, as indicated in the User table. In the
example of Figure, we add a new complexity where users are also given
access to all of their direct reports' DAC permissions, as shown in the Org
table.
By adding a generic join and a self -join on the Org table to locate all of a
user's direct reports, which is then joined to the User table and the
Customer table, SQL queries on the Customer database in a typical
relational database can be adjusted to allow data access control. Joins are
not supported in cloud databases like Google Datastore or Amazon's
SimpleDB, though. As a result, the functionality depicted in the picture
must be implemented in code: The self -join on Org is performed in
memory, providing a list of reportees that includes the user. This list is
used as a filter to obtain the User table's list of allowable DAC IDs.
Finally, the application query on the Custome r table is filtered using this
list.The DAC ID must be recalculated based on the DAC Rules whenever
a customer record is added or modified; this computation must also be
optimised, particularly if there are many DAC Rules. Re -computation and
updates to the DAC ID values in the Customer database are also necessary
for the addition of new DAC Rules or the modification of existing
ones.Additionally, when filling up the DAC Rules table, care must be
given to guarantee that DAC ranges on the same field are never
overlapping.
4.7 SUMMARY
Through their internet -connected PCs, smartphones, tablets, and
wearables, users of cloud computing technologies can access storage, files,
applications, and servers. Data is processed and stored by cloud computing
services away f rom end users.Delivering computing resources through the
internet, such as storage, processing power, databases, networking,
analytics, artificial intelligence, and software applications, is known as
cloud computing (the cloud). In this module we have disc ussed various
technologies related to cloud computing. We have also learnt
fundamentals of cloud computing platforms and infrastructures. We have munotes.in

Page 60

Cloud Computing –II
60 seen the core concepts of web services along with the concept of mashup.
Multi -tenant software concepts are al so introduced.
4.8 LIST OF REFERENCES
1. Enterprise Cloud Computing Technology, Architecture, Applications,
Gautam Shroff, Cambridge University Press, 2010.
2. Mastering In Cloud Computing, Rajkumar Buyya, Christian Vecchiola
And ThamariSelvi S, Tata Mcgraw -Hill Education, 2013.
3. Architecting the Cloud: Design Decisions for Cloud Computing
Service Models (SaaS, PaaS, and IaaS), Michael J. Kavis, Wiley CIO,
2014.
4. Cloud Computing: SaaS, PaaS, IaaS, Virtualization, Business Models,
Mobile, Security and More, Kris J amsa, Jones & Bartlett Learning,
2013.
4.9 UNIT END EXERCISES
1. Define cloud computing and discuss on cloud computing platforms.
2. What do you mean by infrastructure as a service. Explain with the help
of Amazon EC2.
3. Write a note on Google Datastore.
4. Explain A mazon SimpleDB.
5. Discuss on web services.
6. Write a note on SOAP/WSDL web services.
7. Illustrate the concept of REST web services.
8. Compare SOAP Vs REST.
9. Write a note on AJAX.
10. Explain multi -tenant software.
11. Explain multi -schema approach.
12. Write a note on data acc ess control for enterprise applications.


munotes.in

Page 61

61 5
CLOUD TECHNOLOGIES -II
Unit Structure :
5.0 Objectives
5.1 Introduction
5.2 Concurrent computing: Thread programming
5.2.1 What is a thread?
5.2.2 Concurrent execution of threads
5.2.3 Thread states and life cycle
5.2.4 Enhancing Interface Responsiv eness using Threads
5.2.5 Advantages of multithread design
5.2.6 Case study: Cooperating threads
5.3 High -throughput computing: Task programming
5.3.1 Task computing
5.3.2 Characterizing a task
5.3.3 Computing categories
5.3.4 Frameworks for task computin g
5.3.5 Task -based application models
5.3.6 Task programming model
5.3.7 Developing applications with the task models
5.4 Data intensive computing: Map -Reduce programming
5.4.1 What is data intensive computing?
5.4.2 Characterizing data -intensive computat ions
5.4.3 Challenges
5.4.4 Technologies for data -intensive computing
5.5 Summary
5.6 List of References
5.7 Unit End Exercises munotes.in

Page 62

Cloud Computing –II
62 5.0 OBJECTIVES
When you finish reading this chapter, you
 Recognize the idea of a thread
 Know how to create and write applica tions with multiple threads
 be able to use the Runnable interface and the Thread class
 comprehend the thread's lifecycle
 comprehend thread synchronization
 to understand the fundamentals of task and map -reduce programming
5.1 INTRODUCTION
Managing multiple tasks at once is the topic of this chapter. In our
everyday lives, multitasking is a normal occurrence. Let's say, for
illustration purposes, that you had cereal, toast, and a cup of coffee for
breakfast today. Breakfast consists of three simultaneous acti vities: cereal,
toast, and coffee.
Actually, you perform these actions sequentially rather than "at the same
time": You eat some toast, then a bite of toast, and finally a drink of
coffee. You continue eating toast, eating cereal, drinking coffeeis over. I f
there is a call when eating, there might be a situationof picking up & carry
on with your meal —or at the very least, sip your coffee. By doing more
"at the same time," you imply. Numerous instances of multitasking occur
frequently in daily life.
Our prev ious computer programmes only completed singleapplication.
However, at many instances when programme requires performing
numerous tasks simultaneously. For instance, if you created a chat
application for the internet, it would enable multiple users to part icipate in
a discussion. The software needs simultaneously read data and disseminate
them to the other group members. It would be necessary to do the reading
and broadcasting responsibilities simultaneously.
5.2 CONCURRENT COMPUTING: THREAD
PROGRAMMING
Threads and concurrent programming:
Threads can be thought of as methods that run "concurrently" with
other methods. Typically, when writing a computer programme, we think
in sequential fashion. From this angle, just one thing is in motion at once.
However , many processes can really be running simultaneously while
sharing the same memory thanks to modern multi -core computers.
munotes.in

Page 63

Cloud Technologies -II
63 5.2.1 What is a thread?
Executable applications are organised into single thread, often known as a
thread. Imagine making list of t he programs as the computer's central
processing unit executes them in order to visualise a thread (CPU).
Assume splitting a programme into multiple separate threads. Every set of
instructions will be unique. The applicationisperformed sequentially
within a thread, as is customary. Computer can execute many threads
simultaneously. Although CPU only processes single instruction, it may
manage numerous threads at once by rapidly switching between them.
Concurrency's key benefit is that it enables the compute r to perform
multiple tasks at once. The CPU might either download & perform
computation, for instance.From our vantage point, it can appear like the
computer is running multiple CPUs simultaneously, but it is really an
illusion produced by threads that ar e properly scheduled.
5.2.3 Thread states and life cycle
Figure 2 and Table 1 summarise the various stages that each thread can be
in during its life cycle. Labelled ovals denote the different thread states,
while labelled arrows denote the transitions bet ween the states of ready,
running, and sleeping. The Java Virtual Machine and the operating system
are mostly in charge. Those software can control thread transitions
denoted by methodologiese.g.,start(), stop(), wait(), sleep(), and notify().
Because stop ping a thread in the middle of its operation is inherently
risky, JDK 1.2 deprecated the stop() method among these.The CPU
scheduler is in charge of other transitions, including dispatch, I/O request,
I/O done, time expired, and finished sleeping. In ready state when it is first
generated. It is waiting for its turn to run on the CPU. Similar to a waiting
line is a queue. The first thread in the lineis given the CPU, when the CPU
becomes available. When that happens, it will be running.

Figure 1: A demons tration of thread’s life cycle

munotes.in

Page 64

Cloud Computing –II
64 Table 1: Summary of thread’s state

The CPU scheduler, an integral component of the runtime system, controls
the transitions between the ready and running phases. Scheduling
numerous threads effectively and fairly is compa rable to having several
kids share a single bicycle. Ready kids wait in line for their turn to pedal
the bike. The older person (the scheduler) gives the first youngster (the
thread) a few minutes to ride the bike before taking it away and passing it
to th e next person in line. This is comparable to one of the kids choosing
to take a quick break while waiting their turn. The young person would
rejoin the queue after the rest was finished.When a thread uses the wait()
criteria, it leaves CPU voluntarilyuntil some other thread notifies it.
The architecture controls change from blocked state to the ready state. In
comparison to the CPU, I/O devices like disc drives, modems, and
keyboards operate exceedingly slowly. As a result, controllers —separate
processors —handle I/O operations. For instance, the system will send the
disc controller instructions on where to put the data requests. The thread is
halted and another thread is permitted to run because it cannot do anything
until the data are read.
5.2.4 Enhancing Interface Responsiveness using Threads
A multithreaded software can be used to improve user interface
responsiveness. A programme that is running statements in a lengthy
(perhaps indefinite) loop in a single -threaded application does not respond
to user i nput until the loop is ended. As a result, there will be a visible and
occasionally annoying wait between when the user initiates an action and
when the software really handles it.
5.2.5 Advantages of multithread design
Having changed a single -threaded sof tware into a multi -application by
adding a separate thread for Dotty. The drawing task is handled by the
second thread. We ensure that thread will continue to reply to commands
by requiring on sleep in every iteration. Distinction between single and
multit hreaded designs is seen in Figure 3. Be aware that the GUI thread
runs dotty.clear and begins. Simply said, drawing thread runs its draw()
procedure. All of these operations are carried out by a single thread in the
single -threaded version. munotes.in

Page 65

Cloud Technologies -II
65 Drawing N rando m dots will take longer with this design for a brief period
after each iteration. The extra time doesn't really matter, though. The
programme becomes significantly more responsive as a result of splitting
into 2 independent control, 1 handling drawing work & 1
managingapplication interface.

Figure 2: 2 independent threads
5.2.6 Case study: Cooperating threads
The behaviour of threads needs to be synchronised and coordinated in
some applications in order for them to work together to complete a task.
The pr oducer/consumer model serves as the foundation for many
cooperative applications. This approach assumes that two threads work
together to produce and consume a specific information. Consumer thread
reads & uses message or result that the producer thread cr eated. The
producer must be careful of not replacing anoutcomethat’s not utilized,
and the consumer shall await for the outcomes to get generated. The
consumer model is applicable to many different kinds of coordination
issues.
Controlling the visualizatio n of information that is accesses by ourgateway
/ portal is one use case for this architecture. The producer thread writes
information as it comes in from the Internet to a buffer. Information is
read through the buffer and shown in the gatewaytab by a sep arate
consumer thread. The two threads must, of course, be precisely synced.
5.3 HIGH -THROUGHPUT COMPUTING: TASK
PROGRAMMING
The broad field of distributed system programming known as "task
computing" includes numerous techniques for creating distributed
applications.A task depicts a software that needs input files and generates
output files as it runs. Then, applications are made up of a number of
tasks. These are set into action, and when they're finished, the output data
is gathered.This section descri bes the task abstraction and gives a quick
rundown of the distributed application models that are built on it.
5.3.1 Task computing
An operation that produces a unique o/p& can be separated as
auniquerationalentity is referred to as a job. In actuality, a job is expressed
as a separate piece of software, or programmeoperated in a distantdomain. munotes.in

Page 66

Cloud Computing –II
66 The basic goal of multithreaded programming is to facilitate parallelism
on a single processor. By combining the processing capability of
numerous computing nodes, t ask computing offers distribution. As a
result, this paradigm makes the existence of a distributed infrastructure
obvious.Cloud computing has now become a popular way to get a lot of
processing power for the use of dispersedtasks. It requires the right
interface to accomplish. Figure 4 shows a typical task computing reference
scenario.

Figure 3: Task computing scenario
A software layer called middleware allows for the synchronizeapplication
of numerousassets, whether they come from centralizedinformation centre
or a network of PC’s that are spread out geographically. The middleware's
access point(s) receive a user's collection of tasks and handle task
scheduling and task execution monitoring. Every computing tool offers a
suitable runtime environment. The middleware's APIsare used fortasks
submission.
Additionally, suitable APIs are offered for task status monitoring and
result collection after task completion. A group of familiar activities
required by the interface requirements to assess the development a nd
formulation of event -based programs can be identified. The procedures
include:
 Scheduling& coordinating applications for development on a number
of distant junctions
 Managing dependencies and shifting programmes across distant
junctions
 Establishing a s etting for the remote nodes' task execution
 Tracking the progress of every application completion & updating the
customer on it
 Having approach to task's o/p.
5.3.2 Characterizing a task
A task is an application component that can be conceptually isolated and
run independently. Various components can be used to represent a task:
 Shell scripts that combine the execution of many apps
 Only one programme munotes.in

Page 67

Cloud Technologies -II
67  A piece of code that runs in the context of a particular runtime
environment, such as a Java/C11/.NET class
 I/Pfolders, feasible code (programmes, etc.), and O/Papplication define
a task.
Operating system or a comparable sandboxed environment serves as the
tasks' runtime environment. Additionally, a task might require particular
software appliances on the remot e execution nodes.
5.3.3 Computing categories
These classifications offer a broad overview of the traits of the issues.
They unwittingly place demands the middleware & infrastructure.
Applications are classified as:
1] High -performance computing (HPC)
HPC is the application of allocation computer resources to the solution of
computationally intensive problems. A sizable number of computationally
expensive activities that must be completed quickly make up the overall
profile of HPC applications.FLOPS now ter a-FLOPS, or even peta -
FLOPS, which count the nos of FLOPS, are the metrics used to assess
HPC systems.
Example: HPC applications created to address "Grand Challenge"
challenges in science and engineering can be supported by supercomputers
and clusters.
2] High -throughput computing
Using distributed computing resources for applications that need a lot of
processing power over a lengthy period of time is known as high -
throughput computing (HTC). HTC systems must be reliable and sturdy
enough to last for a ver y long time. The typical feature of HTC apps is that
they are composed of numerous activities, each of which can take a long
time to complete.
Example: Statistical or scientific simulations.
Individualistic application that are arranged in remote expedient since they
doesn’t require communication are extremely prevalent. The jobs that are
done each month are how HTC systems gauge their performance.
3] Many -task computing
High -performance computations known as MTC include a number of
unique activities connec ted by bindermanipulations. MTC is diversity of
applicationsconsisting of several types: Compact or big, single or
multiprocessor, compute - or data -intensive, static or dynamic, single or
multiple tasks are all possible. Apps that use the message -passing i nterface
yet are loosely connected and communication -intensive fall under the
category of MTC tasks. munotes.in

Page 68

Cloud Computing –II
68 5.3.4 Frameworks for task computing
The following are some well -known software programmes that assist
framework. Each of thismodel share the overall refere nce architecture
shown in Figure 4 in terms of architecture. A scheduling (single or
multiple) & worker junctions make up their two main parts. The way that
the system's parts are organised can change.
1. Condor
The most popular and durable interface for moni toring groups,
inactivebureau, & a group of agglomeration is called Condor. Having the
ability to control heavily occupied junctionstheysupport the capabilities of
batch -queuing systems. It offers a strong task expediteorganisation that
only organizetask o n expedite with right continuance surrounding. On a
range of resources, Condor can manage both serial and parallel jobs. It is
used to manage infrastructures by hundreds of organisations in business,
government, and academia. A version of Condor called Cond or-G enables
interaction with grid computing resources, including those overseen by
Globus.
2. Globus Toolkit
Grid computing is made possible by the technologies included in the
Globus Toolkit.It offers a complete set of tools for collaborating across
corpora te, institutional, and geographical barriers to manage
expedite&different services. In addition to security and file management,
it also includes s/wfunctionalities, libraries &frameworks for expedite
management& administration. The inter -operational toolk it specifies a set
of interfaces and protocols that allow various parts to collaborate with one
another &exploreexpediteacrossones boundaries.
3. Sun Grid Engine (SGE)
Previously known as SGE, Oracle Grid Engine is interface for managing
distributed resources and workloads. SGE was first created to assist the
execution of jobs on clusters, but after integrating new features, it can
currently handle a variety of resources and functions as middleware for
grid computing. It has extensive scheduling features like budget &
collaborativemanagement, monitoringtasks that have constraint,
frameworks and prior bookings. It allows execution of parallel, serial,
interactive, and parametric activities.
4. BOINC
Grid and volunteer computing are supported via the Berkeley Open
Infrastructure for Network Computing (BOINC) platform. It enables us to
use desktop computers as clientmanipulating junctions used for executing
tasks when they are dormant. Both job check pointing and duplication are
supported.The BOINC server and BOINC cl ient are the two primary parts
of BOINC. The primary node for managing all of the available resources
and assigning tasks is the BOINC server. The firmware element installed munotes.in

Page 69

Cloud Technologies -II
69 on computers & establishes the execution task allocation is known as the
BOINC cli ent. By building computing grids with specialised machines,
these systems are quickly configured in offering muchstagnant support for
task allocation.One needsto select & give their computer's CPU cycles
while installing BOINC clients. The BOINC infrastruc ture is currently
supporting a number of applications, spanning from astronomy and
medicine to astronomy and cryptography.
5. Nimrod/G
Tool for parameter sweep application modelling and execution on large
computational grids. For expressing parametric experim ents, it offers a
straightforward declarative parametric modelling language. It makes use
of cutting -edge expedite monitoring and processingframeworks that are
established. It enables stringent and fund limited application processing on
diversifiedexpedite to reduce processing costs and achieve measurable
outcomes on schedule. Over the years, it has been utilised for a very
diverse range of purposes, from policy and environmental effect to
quantum chemistry.
5.3.5 Task -based application models
Multiple syst ems are built on idea that a task serves as primary building
block for creating diverse applications. The manner in which tasks are
generated, their connections with one another, and the existence of
dependencies or other conditions are what distinguish th ese models from
one another. Based on the task notion, we briefly review the most
prevalent models in this section.
1] Applications that are astonishingly parallel
The type of distributed applications that is the easiest to understand and
consist of embarr assingly parallel programmes. The tasks do not have to
communicate with one another and can be of the same type or of various
types. The majority of distributed computing frameworks support this
class of applications. There is a great deal of flexibility i n how jobs are
scheduled because they are not required to communicate. There is no
requirement that tasks be completed simultaneously; they may be carried
out in any order. These applications' scheduling is streamlined, and it
focuses on assigning tasks to resources in the best way possible. The
Globus Toolkit, BOINC, and Aneka frameworks and tools support
embarrassingly parallel applications.
Rendering of images and videos, evolutionary optimization, and model
forecasting are among the issues. The task of rendering a pixel or a frame,
respectively, is used in both image and video rendering. When using
evolutionary optimization meta heuristics, a job is found by running the
algorithm just once with a specific set of parameters. The same holds true
for applic ations of model forecasting. The majority of embarrassingly
parallel applications come from scientific applications.
munotes.in

Page 70

Cloud Computing –II
70 2] Applications for parameter sweeps
The jobs are identical in nature and the only difference between parameter
sweep apps and other class es of embarrassingly parallel programmes is the
execution parameters. A template job and a set of parameters are used to
identify them. The operations that will be carried out on the remote node
for task execution are defined by the template task. The para meter set
identifies the group of variables whose assignments tailor the template
task to a particular case. The execution of parameter sweep programmes
can be supported by any distributed computing system that offers support
for embarrassingly parallel ap plications.The sole distinction is that all
permissible parameter combinations are iterated over in order to generate
the jobs that will be executed.Aneka offers client -based tools for
visualising creating a template task, setting parameters, and iterating over
all possible combinations. Nimrod/G is naturally built to facilitate the
execution of parameter sweep applications. This applies to a wide variety
of applications. Domain of scientific computing: applications of
computational fluid dynamics, evolutio nary optimization strategies, and
weather forecasting models
3] MPI applications
A standard for creating parallel applications that communicate by sending
messages is called Message Passing Interface (MPI). As a way to bring
together the many distributed s hared memory and message -passing
infrastructures that are available for distributed computing, MPI was first
developed. Today, the de facto industry standard for creating portable and
effective message -passing HPC programmes is MPI.
MPI gives programmers a collection of routines that:
 Manage the distributed environment in which MPI programmes are
executed;
 Provide facilities for point -to-point and group communication;
 Support the definition of data structures and memory allocation;
 Provide basic support fo r synchronisation with blocking calls.
4] Workflow applications with task dependencies
Applications that use workflow are characterised by a group of tasks that
show relationships between one another. These dependencies, which are
typically data dependenci es, determine the manner and location of
scheduling for the applications.A workflow is the whole or partial
automation of a business process in which tasks, information, or
documents are transmitted from one resource (a human or a machine) to
another for a ction in accordance with a predetermined set of rules. The
idea of scientific workflow originated from the concept of workflow,
which can be thought of as the structured execution of tasks that are
interdependent. Workflow has proven to be beneficial for e xpressing many
scientific studies. In scientific workflows, the participants are primarily
computer or storage nodes, and the process is identified by an application
to run. Tasks and data are the most common materials transmitted between munotes.in

Page 71

Cloud Technologies -II
71 participants.A wo rkflow definition scheme that directs the scheduling of
the application defines the collection of procedural rules. Data
management, analysis, simulation, and middleware supporting the
workflow's execution are typically included in a scientific workflow. A
directed acyclic graph (DAG), which specifies the interdependencies
among activities or procedures, is typically used to depict a scientific
workflow. In a workflow application, the nodes on the DAG represent the
tasks that need to be completed; the arcs linking the nodes show which
tasks are dependent on one another and which data channels connect
them.Data dependency, which occurs when the output files of one activity
serve as the input files for another task, is the most prevalent type of
dependency tha t may be achieved through a DAG.
5.3.7 Developing applications with the task models
Several components are required for task -based programme
execution.Such apps can only be created via the following actions:
 Creating classes that use the ITask interface
 Creating an AnekaApplication instance that is appropriately
configured
 Creating AnekaTask objects and wrapping ITask instances around
them
 Running the programme and watching for it to finish
5.4 DATA INTENSIVE COMPUTING: MAP -REDUCE
PROGRAMMING
The focus of d ata-intensive computing is on a group of applications that
work with a lot of data. Large amounts of data are generated by a variety
of application domains, from computational science to social networking,
and these volumes need to be effectively stored, a ccessed, indexed, and
evaluated. These activities become more difficult as knowledge gathers
and grows at a faster rate over time. In order to overcome these issues,
distributed computing offers more scalable and effective storage designs
as well as improv ed data computation and processing capabilities.
5.4.1 What is data intensive computing?
Production, manipulation, and analysis of massive amounts of data,
ranging from hundreds of megabytes (MB) to petabytes (PB) and beyond,
are the focus of data -intensiv e computing. A collection of information
pieces that are pertinent to one or more applications is frequently referred
to as a dataset. Datasets are frequently kept up -to-date in repositories,
which are infrastructures for storing, retrieving, and indexing massive
amounts of data. Relevant data points, known as metadata, are attached to
datasets to aid with classification and search.
Numerous application domains involve data -intensive processing. The
most well -liked one is computational science. Many times, those who run
scientific simulations and experiments are eager to generate, examine, and munotes.in

Page 72

Cloud Computing –II
72 interpret enormous amounts of data. Every second, telescopes mapping the
sky generate hundreds of gigabytes of data; over the course of a year, the
collection of these photos easily exceeds a petabyte in size. Applications
for bioinformatics mine databases that could ultimately hold terabytes of
data. Massive amounts of data are processed by earthquake simulators as a
result of the Earth's tremors being recorded all aro und the world.
5.4.2 Characterizing data -intensive computations
Applications that deal with large amounts of data frequently also have
computationally intensive characteristics. The two top quadrants of the
graph are where data -intensive computing is foun d, according to Figure 5.
Datasets that are many terabytes and petabytes in size are handled by data -
intensive applications. Datasets are frequently dispersed across various
sites and stored in a variety of formats. These applications use multi -stage
analy tical pipelines to process data, including transformation and fusion
steps. The processing demands increase practically linearly with the
amount of data, and parallel processing is simple. They also require
effective data management, filtering, and fusion, as well as effective
querying and delivery systems.

Figure 4: Data -intensive research issues
5.4.3 Challenges
The enormous volume of data generated, processed, or stored places
demands on the middleware and supporting infrastructures that are rarely
met by conventional distributed computing solutions. For instance, the
location of data is essential because high -performance computations must
avoid having to move terabytes of data. Scalable algorithms, content
replication, and data partitioning all aid in enhancing the performance of
data-intensive applications.
munotes.in

Page 73

Cloud Technologies -II
73 There are still open challenges in data -intensive computing:
 Massive datasets can be searched and processed using scalable
algorithms
 Adaptable new methods for managing metadata that can handle
complicated, heterogeneous, and remote data sources
 Platform improvements for high -performance computing that attempt
to improve access to in -memory multitera -byte data structures
 Petascale, high -performance, and dependable distributed file systems
 Data signa ture-generation methods for quick and efficient data
processing
 There are new methods for software mobility that can give algorithms
that can transfer computation to the location of the data
 Specialised hybrid interconnection topologies that better enable
filtering multigigabyte datastreams from fast networks and scientific
equipment
 Techniques for combining software modules running on several
platforms to quickly assemble analytical pipelines that are adaptable
and performant
5.4.4 Technologies for data -intensive computing
The creation of applications that are primarily concerned with processing
massive amounts of data is known as data -intensive computing. Thus, the
technologies enabling data -intensive computing can be naturally
categorised as storage syste ms and programming models.
1] Storage systems
Database management systems have historically served as the de facto
storage support for a variety of application kinds. The relational model in
its original form does not appear to be the preferable method for
supporting data analytics on a wide scale due to the proliferation of
unstructured data in the form of blogs, Web pages, programme logs, and
sensor readings. There are fresh chances as database research and the data
management sector reach a turning point . This shift is caused by a number
of things, including:
 Growing interest in big data
A number of industries, including scientific computing, enterprise
applications, media entertainment, natural language processing, and social
network analysis, now routin ely manage massive amounts of data. New
and more effective strategies for data management are required due to the
huge amount of data.
 The increasing significance of data analytics in the supply chain
Data management is now seen as a crucial component of b usiness profit
rather than a cost. This problem occurs in well -known social networks like
Facebook, which rely on the administration of user profiles, interests, and
connections between people. In order to support data analytics, the munotes.in

Page 74

Cloud Computing –II
74 enormous volume of dat a that is constantly mined necessitates new
technologies and approaches.
 Presence of data in formats other than just structured
As was already established, today's relevant information is diverse and
comes in a variety of shapes and formats. The use of tra ditional enterprise
applications and systems has resulted in the growth of structured data, but
at the same time, technological advancements and the democratisation of
the Internet as a platform where anyone can access information have
produced a vast amou nt of unstructured data that does not naturally fit into
the relational model.
 Modern methods and technology in computing
Access to a huge quantity of processing power is promised by cloud
computing. As a result, software engineers can create systems that scale
progressively to any level of parallelism. Building software that is
dynamically deployed on hundreds or thousands of nodes, some of which
may be part of the system for a few hours or days, is no longer unusual.
Traditional database infrastructures a re not made to sustain such an
unstable environment.
These elements collectively point to the demand for innovative data
management technology. This not only suggests a fresh research agenda
for database technology and a more all -encompassing method of
information management, but it also creates potential for rival (or
complementary) models to the relational model. The main prospects for
supporting data -intensive computing include, in particular, developments
in distributed file systems for the administrati on of raw data in the form of
files, distributed object stores, and the expansion of the NoSQL
movement.
5.5 SUMMARY
In this chapter we have discussed about various aspects of computing such
as concurrent computing, high -throughput computing and data inten sive
computing. Concurrency refers to the simultaneous execution of several
calculations. Programming in the present day is rife with concurrency. We
have also discussed about the task abstraction and gives a quick rundown
of the distributed application mo dels that are built on it.
The primary characteristics of data -intensive computing were covered in
this chapter. High volumes of data are processed or produced by data -
intensive applications, which can also display compute -intensive
characteristics.Along w ith the technology, programming languages, and
storage structures utilized for data -intensive computing, the data volumes
that led to the definition of data -intensive computation have varied
throughout time. The area of data -intensive computing first gaine d
popularity in high -speed WAN applications. munotes.in

Page 75

Cloud Technologies -II
75 The term "Big Data" currently refers to data dimensions that exceed the
size of terabytes or even petabytes and are stored in storage clouds. This
phrase refers to the enormous amount of data that is generated, analyzed,
and mined by organizations that offer Internet services like search, online
marketing, social media, and social networking, in addition to scientific
applications.
5.6 LIST OF REFERENCES
1. Enterprise Cloud Computing Technology, Architecture, Appli cations,
Gautam Shroff, Cambridge University Press, 2010.
2. Mastering In Cloud Computing, Rajkumar Buyya, Christian Vecchiola
And ThamariSelvi S, Tata Mcgraw -Hill Education, 2013.
3. Architecting the Cloud: Design Decisions for Cloud Computing
Service Models (S aaS, PaaS, and IaaS), Michael J. Kavis, Wiley CIO,
2014.
4. Cloud Computing: SaaS, PaaS, IaaS, Virtualization, Business Models,
Mobile, Security and More, Kris Jamsa, Jones & Bartlett Learning,
2013.
5.7 UNIT END EXERCISES
1. What is a thread?
2. Explain the proce ss of concurrent execution of threads.
3. Explain thread states and its lifecycle.
4. State advantages of multithread design.
5. Define and explain task computing.
6. Discuss on characterizing a task.
7. State the framework for task computing.
8. Write a note on task -based application models.
9. Explain task programming model.
10. What is data intensive computing?
11. Describe characterization of data -intensive computations.
12. State the challenges of data intensive computing.
13. Describe the technologies for data -intensive computation.
 munotes.in

Page 76

76 6
SOFTWARE ARCHITECTURE -I
Unit Structure :
6.0 Objectives
6.1 Introduction
6.2 Dev 2.0 platforms
6.2.1 SALESFORCE.COM’S FORCE.COM Platform
6.2.2 TCS instant apps on Amazon cloud
6.2.3 More Dev 2.0 platforms and related efforts
6.2.4 Advantages, Applic ability and disadvantages of DEV 2.0
6.3 Enterprise software: ERP, SCM, CRM
6.3.1 Anatomy of a large enterprise
6.3.2 Partners: People and Organizations
6.3.3 Products
6.3.4 Orders: Sales and Purchases
6.3.5 Execution: Tracking work
6.3.6 Billing
6.3.7 A ccounting
6.3.8 Enterprise processes, build Vs buy and SAAS
6.4 Summary
6.5 List of References
6.6 Unit End Exercises
6.0 OBJECTIVES
 To understand the various distributed system models and evolving
computing paradigms
 To gain knowledge in virtualization of computer resources
 To realize the reasons for migrating into cloud
 To introduce the various levels of services that can be achieved by a
cloud
 To describe the security aspects in cloud and the services offered by a
cloud
munotes.in

Page 77

Software Architecture -I
77 6.1 INTRODUCTION
The scale th at is achievable in a massivediversifiedsituation where H/W&
network problems can arise more frequently than not has been the focus of
the cloud development paradigms.
This class of applications is technically referred to as "forms -based
transaction -proces sing" but it is constrained by scalability, so we omit
those that must support extremely high transaction and data volumes.
There have been various attempts over the years to create common
abstractions to represent such systems in order to make them more e asily
constructed.
6.2 DEV 2.0 PLATFORMS
Now, we focus on commercialusage, which doesn’t need high amount of
processing but make up a significant portion of software development
efforts: Enterprise IT departments devote a lot of development and
maintenance time to the multiple inter commercialframework. This class
of applications often has computing needs that are orders lower than those
of target critical applicationor massive visualization tasks. Due to vast
number of such systems in operation, these appl ications represent a
significant class of applications.
6.2.1 SALESFORCE.COM’S FORCE.COM Platform
It hosted multiple tenant customer relationship management (CRM)
system was among first successful software as a service products. One of
the factors for the hosted model's appeal was the ability for highly mobile
salespeople to access CRM data online.However, the configurability of
Salesforce.com's CRM system played a more significant role in its growth.
The main CRM product's screens have always allowed end u sers to add
custom fields with ease; these fields are automatically uploaded to the
database, but only for the specific client who created them. Additionally,
as a configuration setting, this process might be carried out while utilising
the CRM software fr om a web browser.
here was no need for programming or a different development tool.
Similar to this, straightforward frameworks can be implemented, which
incorporated in tasks added to users' to -do tasksdepending on the
development & update of information meeting specific criteria: For
instance, a head may be alerted during a salesperson filed an opportunity
with a value higher than a predetermined threshold. Through
configuration, end users were able to modify CRM model to suit their
requirements without a ny help. We illustrate this with an example below
as shown in figure 1:
Consider an application for employee services that keeps track of each
employee's name, contact information, &holidayremaining, or the no of
unused vacation. The application enables t he creation of Leave Request
(LR) records. Additionally,computation must be done when submitting a munotes.in

Page 78

Cloud Computing –II
78 new vacationapplication that rejects the request. However, if there is
enough leave remaining, the request is approved and the amount of leave
requested is s ubtracted from the leave balance.

Figure 1: Leave Request example
This straightforward illustration offers functions that address several
requirements for fundamental forms -based office automation: forms that
can link fields from one application to diffe rent one, etc. are examples of
forms with fields of many sorts (including dates). Thisserves as an
example of Dev 2.0, demonstrating that a somewhat complicated task may
organized together instead of created using straightforward webportal
rather than a pr ogramming environment. Such configuration is made
possible by Dev 2.0 platforms in a user -friendly way.
Figure 2 illustrates few of actions required to create the LR application
using Force.com, e.g., incorporating a certain change. This entails
numerous p hases during which different information needs to be
incorporated and addressed&they may accomplish using the application.

Figure 2: Adding fields in Force.com
The LRapplication generated using aforementioned procedures is shown
in Figure 3, along with the APEX code needed to carry out the server -side
calculations necessary when a contemporary LR is developed. This creates
a "set off" on the LR that will run the moment a record of this type is
entered. They use SQL to approach the directory index&improvi se data in munotes.in

Page 79

Software Architecture -I
79 the index. You can conduct pretty general computations by uploading
code that uses triggers. APEX development, on the other hand, is a task
for programmers as opposed to end users, and is typically carried out in
Eclipse.The platform stops being a tool that end users can customise and
starts functioning as a hosted development environment for programmers.

Figure 3: LRwith APEX application
6.2.2 TCS instant apps on Amazon cloud
The software as a service (SaaS) paradigm is the foundation provided by
Force.com. In this approach, estimation and informationincorporating on
firmware&executing on servers owned & maintained by the SaaS
provider. End users restrict ingress to and visibility that houses the
information. Additionally, the SaaS platform pro vider is solely responsible
for ensuring that their apps perform as demand rises: Users are unable or
unwilling to manage resources allotted in advance of changes in
demand.Here, we outline an alternative strategy used by the InstantApps,
which was created by the research team at TCS R&D.
It is packaged like many other development tools and made accessible on
EC2 infrastructure. The deployment model is similar to many aspects of a
SaaS product, such as a method that allows customers to receive regular
upgra des and also forces them to do so. This is in contrast to SaaS models,
which do upgrades in a way that is invisible to the end user.
Figure 4 compares the deployment and use of InstantApps on Amazon
framework to more conventional Dev 2.0 offerings. Here it only permits
apps to run against production data; the InstantApps Designer AMI
includes tools that enable application development and customization. The
user retains control over the size and number of servers to boot, as well as munotes.in

Page 80

Cloud Computing –II
80 the production data, in t his way.Users can connect other programmes to
their data or use SQL to access their production data.

Figure 4: InstantApps on Amazon EC2
The InstantApps Designer establishes a connection with one of the Model
Repository servers; as designers build their applications, this repository is
filled with application meta -data. Additionally, customers are advised of
newer InstantApps versions via this link, and when such upgrades occur,
which can happen very frequently, exactly like in a SaaS model.
This will pr otect production servers from upgrades until they are
genuinely needed. As a result, when it comes to upgrades, the cloud -based
InstantApps deployment strategy benefits from both SaaS and
conventional installed software solutions.
6.2.3 More Dev 2.0 platfo rms and related efforts
The FADS architecture, which was created for a VAX VMS platform but
reflects insufficient exploration in the domain, is a platform that is quite
similar to contemporary Dev 2.0 platforms! Later, the scholarly PICASSO
framework & ABF tool also made use of a similar methodology. In
contrast to the interpretative, multi -tenant designs of Dev 2.0 platforms,
these are what we refer to as "first -generation" forms interpreters.
Similar to Force.com, there are now other web -hosted Dev 2.0 pl atforms
available from new businesses. Some of these have either shut down, like
Coghead, accessed by hugebusinesses, including Jotspot&Nsite..Each of
these platforms supports the development of straightforward form -based
workflows and is built on an inter pretive, multi -tenant architecture.
Similar to Force.com's APEX, many also provide the scripting of server -
side business logic. Coghead offered a comparable and more platforms
will undoubtedly follow suit in the future.
There have also different visualizat ion editors and interpretative runtimes
put forth in literature and on the market: WebRB that describes online
workframe that work with entities and is well -suited for graphical editing
and visual display. Statesoft is a for -profit tool that, like WebRB, m odels
user interface interactions using state -charts.
munotes.in

Page 81

Software Architecture -I
81 6.2.4 Pros, Applications and disadvantages
Small - to medium -sized business application development could undergo
a paradigm shift thanks to Dev 2.0 platforms, which could also result in
considerable ga ins in efficiency. Their accessibility as cloud services
makes it easier for end users to use them, overcoming any potential
resistance from corporate IT. However, the types of apps that may be
developed using the Dev 2.0 paradigm and the interpretive meth od are
constrained.
We first look at the reasons why developing web -based applications is
challenging. A distributed design that demands a multiple layerdesign
format and has an impact on the software process, particularly
downstream, adds to the complexit y of forms -based application
development (coding and test). The technology needed for each tier varies,
such as writing HTML pages etc. As a result, numerous developers are
typically (rather than occasionally) into creation of every entity.
The deployment stage can be skipped if it is known that the code
generators in each tier produce code that is (a) accurate and (b) correctly
interacts with code produced in other layers. However, it is impossible to
completely avoid the deployment, reinitializing, and re starting steps.
An alternate strategy is the interpretative Dev 2.0 architecture.Code
generation does speed up development, however Dev 2.0 is more
effective: Dev 2.0's "create and immediately play" methodology enables
functional changes to be performed wh ile an application is already
running, allowing for immediate testing of the modification. The
modifications made by other developers are immediately visible to each
developer.
All usual benefits of fast prototyping platforms, such as decreased cycle
times and decreased errors as a result of better requirements elicitation, are
now available with WYSIWYG application development in platforms like
InstantApps.
It allows important entities to be generated by client rather than coders
multiplies all these ad vantages. Access is made possible without any
infrastructure investments thanks to a cloud -based deployment
architecture. Finally, the hybrid cloud deployment architecture used by
InstantApps allays the frequently voiced worry about data ownership.
Despite all of this, the Dev 2.0 model has several drawbacks. First off, in
our experience, there are inherent limitations to how well a method
incorporates the functionality required by modest commercial tasks.
In a scripting language (which is sometimes propri etary), to execute
processing logic. The functionality that can be modelled those seen in
older versions is still constrained. For instance, file I/O and complex string
processing cannot currently be implemented using such models. Apps also
adhere to few p redefined patterns. Since fundamentally altering these
patterns is either difficult or necessitates scripting in JavaScript or another munotes.in

Page 82

Cloud Computing –II
82 proprietary language, the end -user still has no influence over this level of
customisation.
Finally, we haven't seen any extremely complicated applications using
Dev 2.0 yet, such ones with thousands of panels and hundreds of
tables.The WYSIWYG approach's seeming simplicity might actually
result in increased complexity.
6.3 ENTERPRISE SOFTWARE: ERP, SCM, CRM
We considered en terprise designs primarily from a technical standpoint,
charting their development burgeoning concepts.Now, we'll concentrate
on business apps and the tasks they carry out within a huge organisation.
We focus on the data kept in cross -enterprise programmes including
enterprise resource planning (ERP), supply chain management (SCM), and
customer relationship management (CRM).
6.3.1 Anatomy of a large enterprise
First task is to extract the structure of a huge firm, regardless of the
industry to which it may belong. In actuality, this illustration also applies
to organisations focus on corporations here for the sake of clarity. We start
by thinking about the manufacturing of MRP.
Consequently, what does a manufacturing company do? It schedules what
things to produce, when, and how much. The company then carries out the
product process, manages production among factories, give it to clients, &
offers service once they purchased the by -product. In addition to
manufacturing, these fundamental activities are also prevalent in other
industries.
Since they are referred to as "information" systems, corporate applications
must, in the first place, maintain track of data pertaining to business
activities. By creating and modifying this information, the primary
busines s operations are then controlled. Our illustration for a company is
shown in Figure 5.

Figure 5: Core enterprise data munotes.in

Page 83

Software Architecture -I
83 An organisation has customers, and it uses marketing and sales activities
to connect with them. These clients put orders for merchandise . The
company then ships products to clients after delivering them from an
inventory. Keep in mind that if an organization's business model is to
"create for stock," then products are stocked as inventory.
Executing a procedure —or, more generally, putting out some effort —is
the method by which a product is manufactured. Together with other
necessary material inputs, the cost of such activity needs to be monitored.
Additionally, certain material inputs must be purchased. Finally, bills are
sent to clients a nd post -sale services are performed. To determine the
corporation's profit, both incoming revenue and incurred costs must be
taken into account. Along with managing the material assets possessed
and the energy used, the organisation also needs to manage th e people
(human resources) within it.
All of the aforementioned information requirements are equally relevant to
manufacturing and other industries, but with appropriately modified
semantics. Manufacturing execution systems, or MES, or detailed
information tracking, analysis, and optimization, have recently focused on
processes in context of manufacturing industries. Last but not least, the
company adds strategic value by organising and controlling all action
planswith the help of budgets.
The first widespr ead cross -industry applications were manufacturing MRP
systems. Later, these systems were known as ERP, or "enterprise resource
planning" systems, because they extended beyond just manufacturing to
include accounting, sales, and suppliers. Eventually, the roles of the
customer and supplier grew more intricate and started to assume distinct
identities.
As shown in Figure, there are imbrications in categorization. For example,
order administration&payment, which have significant involvement but
are typically categorised under SCM and ERP, respectively, share many
similarities and interact with one another. Additionally, there are other
situations in which packaged solutions might not be appropriate, as well as
particularly unique processes that require softwa re that has been
specifically designed for them.
6.3.2 Partners: People and Organizations
Corporations must interact with other businesses, such as their suppliers or
consumers. Both the employees of these firms and the people who buy
their products must b e dealt with. Early information systems frequently
used different applications to keep track of customers, suppliers, and
personnel individually. Important details could thus frequently be missed
during routine operations; for instance, a customer complain ing who also
happened to be a senior executive of a significant corporate client would
not receive special treatment from a call centre staff. Additionally, there
were missed chances to upsell and cross -sell to current clients. munotes.in

Page 84

Cloud Computing –II
84 Nowadays, it's standard prac tise to combine customer, supplier, and
employee model whenever possible. This entails portraying individuals
and organisations as "partners," each of whom may play a variety of roles.
A straightforward partner model is shown as a UML class diagram in
Figure 6. Since either a person or an organisation might be a partner, this
relationship is generalised in the UML by using class inheritance. In this
relationship, partners may take on one or more partner responsibilities, and
it is portrayed as an aggregatio n.Through a role hierarchy, the partner role
is also subclassified into other real roles, like that of a client, supplier, or
employee.The PARTNER table and PARTNER -ROLE table, with partner
ID serving as a foreign key to indicate a one -to-many link between these
tables, are a partial representation of this class model in terms of relational
tables, as illustrated in Figure. Notably, we have not demonstrated how the
role hierarchy could be implemented relationally and how that would
affect the PARTNER -ROLE t able for the sake of simplicity. The process
of converting a logical class model into a relational representation
involves a number of steps, including several methods for converting class
inheritance to relational tables.

Figure 6: Partner model
As show n in Figure 7, communication with partners occurs through
established contact methods, which might be postal or electronic. As
before, a partial relational representation of this model is shown, in which
the Partner -ID foreign key in the COMMUNICATION -EVEN T table is
used to express the one -to-many link between the partner and
communication event classes. Additionally, this table's Communication -
In-Partner -Role property denormalizes and captures the many -to-many
relationship between communication event and p artner role.Usually,
communications also have additional aspects of the organisation model
attached to them, such as a "work order" for billing any charges incurred
in order to carry out the communication. munotes.in

Page 85

Software Architecture -I
85

Figure 7: Partner communication model
We have gr eatly streamlined the partner model. The internal
organisational structure of organisations, their staff hierarchies, the history
of communications preserved as threads, and many more notions are
commonly included in a real enterprise's highly complicated model.
6.3.3 Products
As seen in the model in Figure 8, an organisation both creates and
consumes products, which might be either goods or services. A business
frequently produces or consumes a large number of items, making it
crucial to categorise them in a variety of ways, such as by model, grade,
segment, or other product category. Product categories may contain or be
contained within other categories; for instance, the categories "paper" and
"office supplies" both include pencils and paper. Take note of how a UML
association class is used to depict a many -to-many relationship like this.
This would be equivalent to a "link table" in a normalised relational data
model with the product -ID and category -ID as foreign keys pointing to the
PRODUCT and PRODUCT -CATEGORY tables.

Figure 8: Product model munotes.in

Page 86

Cloud Computing –II
86 One or more features may be present in a given product. The price of a
product can vary depending on these aspects, the product itself, or other
order parameters like the value or type (for example, retail vs. bulk ) of an
order.
Models of genuine products and their supply are shown in Figure 8.
Physical commodities are preserved as inventory items, which are kept in
storage facilities like warehouses or manufacturing plants, as opposed to
services. Inventory items t hat aren't made by the company itself must be
bought from some supplier businesses. Suppliers of these products are
tracked, along with their ratings, preferences, and contract prices.
Additionally, details on the products that a specific provider is capab le of
creating are also recorded. Finally, reorder criteria for each product set
policies on how much inventory is to be retained.
6.3.4 Orders: Sales and Purchases
In order to conduct business, a firm must first take sales orders before
shipping products or rendering services in response to them. Additionally,
it makes procurements, or orders for goods that it could need. Both sorts
of orders require tracking of many of the same factors, including who
placed the order, where and when it is to be delivered, at what price and in
comparison, to what quotation, how the order was placed, and who gave
their approval.
A combined model for sales and purchase orders is shown in Figure 9. A
number of order elements make up each order. A product or a product
character istic may be the subject of each order item. Each order involves
multiple parties performing various roles, including who placed the order,
who will be billed, and who will receive each order item. Additionally,
orders are placed using a contact method pro vided by the relevant partner,
and each order item includes the contact method for shipping
requirements.Several shipment elements are included in shipments that are
made to fulfil orders. It is crucial to remember that a shipment item may
end up fulfillin g several order items in perhaps distinct orders, or vice
versa, a shipment item may satisfy numerous order items in one order. As
a result, a class association order shipment that tracks the quantity in a
specific shipment item against a certain order ite m connects order items
and shipment items. munotes.in

Page 87

Software Architecture -I
87

Figure 9: Order model
A supply chain application typically handles the order model shown in
Figure 9. However, the sales process, which is often managed by a CRM
application, is responsible for moving a client r equest from a request to a
sales order. The procurement element of SCM typically includes the
corresponding process for purchasing. In either scenario, a request is often
made to one or more suppliers or received from the client prior to placing
an order.F igure 10 shows a quote model that uses quotations and
agreements to follow inquiries to orders: Many request items may be
included in a request. The responding company provides a quote that
includes quote items, each of which is for some products or servic es that
the company can offer for a given price. Following the exchange of
quotes, negotiations take place, and a contract (or agreement) is eventually
drafted between the two businesses. The agreement specifies the pricing
structure, which consists of sev eral agreement items that correspond to
price components for the requested products, including any reductions and
terms that have been agreed upon. The identical pricing components found
in the agreement items are then used to base orders against the quote .

Figure 10: Quote model munotes.in

Page 88

Cloud Computing –II
88 6.3.5 Execution: Tracking work
In addition to providing commodities, businesses also offer services that
necessitate the use of their staff's labour and may be billed either directly
or indirectly. Additionally, costs linked with product design and
manufacturing, R&D, and other such internal operations must be
documented. Figure 11's architecture keeps track of work orders, which
might be generated in reaction to incoming sales orders or internal
needs.Note that we have used multi ple inheritance to illustrate the reality
that work orders may occur in two separate ways. Additionally, this
implies that our model does not distinguish between a work order and a
sales order or an internal requirement. Work instructions frequently go by
the moniker "projects" and have specific names. Thus, our methodology
guarantees that a project to assure an order's completion is likewise started
as soon as an order is received (or an internal requirement is accepted).

Figure 11: Work model
By putting forth effort to complete work orders, the specifics of the
intended and executed efforts are recorded. As illustrated by the effort
breakdown class, efforts can be divided into smaller efforts. It should be
noted that a work order can be completed by nume rous efforts, and the
model permits an effort to contribute to multiple work orders. For
simplicity's sake, a work order is typically connected with a single effort
in practise.
Our work model accurately depicts the key elements of effort tracking,
such as how a need leads to work orders that may involve a complicated
series of planned and completed efforts. It does not, however, include
additional costs related to completing a work order, such as the usage of
tangible assets, consumables, and travel costs. Furthermore, internal
requirements that result from planning can be fairly complex because they
might arise for a variety of reasons, including projected sales, scheduled
maintenance, sales and marketing initiatives, or research and development,
to which linkages may need to be maintained. munotes.in

Page 89

Software Architecture -I
89 People who are working on their given jobs expend effort. The party
assigned to a specific effort is depicted in the model in Figure 12, where
parties might be either organisation personnel or outside contractors. An
associated rate may be assigned to an effort when it is assigned to a party.
Alternately, the rate (cost or fee) might be determined by the pay scale of
a particular employee or contractor. The rate class includes the actual cost -
rate or billed -rate. Each rat e may have a corresponding rate -type, often
known as a rate "schedule."The rate schedule to use for a specific work
order is also recorded in the work order class, as was previously
demonstrated in Figure 11, allowing for the use of the proper charging or
costing rates while completing the work order.

Figure 12: Rates and timesheets
Once parties have finished their work, they record it in a timesheet that
contains timesheet elements. Each timesheet item represents the amount of
time devoted to a specific work task. Timesheets are essential for
invoicing, costing, and accounting in many areas of the service sector as
well as shop floor activities in industrial facilities.
6.3.6 Billing
Customers must be billed and payments must be made after things have
been delivered. Similar to this, payments must be made for supplier
purchases. In either scenario, payments must be linked back to the
invoices that billed for the billable goods or services for which they were
issued. A fundamental billing structure is shown in Figure 13. Invoices
can be sales invoices or purchase invoices because billing entails sending
an invoice to a partner (or receiving one from a supplier). We incorporate
all of them via a class association invoice role between invoice and partner
since there may be more partners linked with an invoice, such as the "ship
to" or "ordered by" partner, which may differ from the "bill to" partner. munotes.in

Page 90

Cloud Computing –II
90

Figure 13: Billing model
An invoice may have a number of invoice items, each of which includes
the quantity an d description of the products or services being billed as well
as the total price due. Each invoice item may relate to a particular
inventory item, a product feature, or both; these are consequently
displayed as optional linkages to the respective classes. In addition, each
line item on an invoice is associated with the actual delivery of some
products or services, i.e., one or more billed items. A billed item may be a
shipment item, an effort, or a timesheet entry. Invoices can also be
invoiced on order; i n which case a billed item might be an order item.
Payments are credited to an invoice once it's been sent. (For purchases,
this is the other way around; payments are made against invoices
received.) A payment model is displayed in Figure 14. Payments are made
by a partner and might take the form of receipts or disbursements. In a
payment, the sum being paid, a reference number, and the day the
payment becomes effective are all recorded (as opposed to when it is
physically received). One or more invoice ite ms may be used as the
recipient of a payment, and several payments may be made to one invoice
item. This many -to-many link is modelled as a class association for
payment invoices.Last but not least, every payment triggers a financial
transaction in a bank account; revenues trigger deposits, and payments
made result in withdrawals. Keep in mind that financial transactions
between legal entities occur in the actual world through a third -party
financial institution (such as a bank). munotes.in

Page 91

Software Architecture -I
91

Figure 14: Payment model
6.3.7 Accounting
The financial situation of an organisation is undoubtedly impacted by
financial transactions like payments. But so are unpaid invoices that have
been raised, as well as inventory accumulation that could result in a loss if
it cannot be sol d.The accounting model of the business captures the
financial position of the company, and accounting transactions are made to
reflect the effects of each business activity (such as actual payments or
simple invoices) on the company's financial position.
The chart of accounts, which consists of a collection of "general ledger"
(GL) accounts, serves as a visual representation of an organization's
financial structure. Each GL account serves as a container for posting
accounting transactions.Figure 15 shows a straightforward accounting
approach. Each account in the general ledger has a GL -ID, a name, and a
description. Examples include "travel expenses," "accounts receivable,"
"cash" accounts, and "payments owed."A GL account also has an account -
type that descr ibes where it belongs on the balance sheet of the company,
indicating whether it monitors assets, liabilities, revenues, or expenses.

Figure 15: Accounting model munotes.in

Page 92

Cloud Computing –II
92 The enterprise's method for capturing financial data is determined by the
chart of accounts, which consists of the set of GL accounts. A large
organisation, however, requires financial reporting at the level of each
budget entity, or an entity with the capacity to spend or receive money,
such as a manufacturing facility or project. Chargeable acc ounts, against
which actual accounting transactions are posted, connect GL accounts to
such budget entities.A profit and loss statement or comprehensive balance
sheet can be created for any chargeable account or collection thereof,
including the organisati on as a whole, using chargeable accounts and the
relationships among them (not shown in the illustration).
As was already indicated, certain business transactions are recorded as
accounting transactions. An accounting transaction can be both internal
and e xternal, such as a transfer charge between two divisions of the
company. Such external transactions may involve actual money
exchanges, such payments, and are thus recorded in a GL account
designated as "cash." Alternatively, these could be invoices that a re sent to
an "asset" GL account because they represent a future obligation by
another party to pay. When an invoice is unpaid, it turns into a bad debt
and must be moved to another GL account designated as a "liability" by
matching credit and debit accoun ting transactions.
Even from the incredibly simplified model above, it is clear that
accounting is a complicated function. It is also a fundamental role of any
enterprise's IT systems. Since generic accounting systems are provided via
packaged software, ac counting functions across businesses are quite
similar. As a way to automate the capturing of business transactions and
tie them to accounting transactions in a flexible way so that one system
could meet the demands of a company, the first ERP systems emer ged
from such accounting packages. This is both a benefit and a drawback of
integrated ERP programmes.
6.3.8 Enterprise processes, build Vs buy and SAAS
The accounting function was the focus of the initial ERP systems, which
aimed to broaden it to include various company operations. Users would
have access to various business transaction kinds depending on their
positions. As a result, salespeople would have access to the production of
orders and the preparation of invoices but not to the tracking of effort s or
shipments. In such a case, it was difficult to follow an order from its start
until payment was collected. For a company to operate effectively, it is
crucial to track such an "order to cash" process. Managing inventory or
procurements involves simila r end -to-end procedures. Each of these
"horizontal" processes typically spans many areas of the company data
model.
An organisation should ideally implement end -to-end operations on a
uniform underlying data model if it were to construct its systems from t he
ground up.However, using packaged software makes the problem more
challenging. Consider the scenario where a core ERP package, for
instance, does not track shipments and specialised software is created to munotes.in

Page 93

Software Architecture -I
93 do so. We have already seen the necessity to cor relate shipments to order
items. The order items must match those kept in the primary ERP system.
Using the different APIs that a core ERP system typically provides; all of
this is still possible. Now imagine that an SCM and a core ERP system are
the two k ey software applications being used.Additionally, order items
and customer shipping information will be maintained by the SCM and
synchronised with the primary ERP system. Customer information is now
being kept in three systems that require constant synchr onisation after the
addition of a CRM package.
Domain -specific functionality causes additional issues, suchcompared to
the cross -industry roles we have discussed thus far, for example:
Organizations providing financial services could have to deal with trad ing
and securities and transactions. These demand a sophisticated partner
model with brokers, clients, stock exchanges, and banks among others. A
typical CRM package would includeunlikely to include these components.
At the same time, it would be costly to redevelop the essential CRM
functionality included in a CRM package. Thus, decisions regarding
"build vs. buy" must take integration issues into accountbetween software
programmes, old computer systems, and unique applicationscan be
contrasted with the su bstantial expenses associated with creating and
sustaining a singlea combined application using a single data model.
When choosing which enterprise applications are suitable for software as a
service, additional factors come into play (SaaS). In this case, in addition
to data security concerns (or worries), it is crucial to take into account both
the volume of data that would need to be routinely transmitted from the
SaaS system to applications implemented on -premises as well as the
degree of data replicati on. CRM systems are advantageous for SaaS from
this perspective because the only point of replication is certain client
information. As new customers are acquired and customer information is
updated more gradually than during business transactions, there i s less of a
burden on the SaaS product to synchronise client data with on -premise
applications.
Another obvious SaaS contender is order management, particularly when
web-based orders are involved. Another instance where SaaS is a viable
choice is in human resources: A SaaS HR system may provide a monthly
payroll and even facilitate direct debits and credits from the organization's
and employees' bank accounts. However, if SCM or other fundamental
ERP components are taken into account, the scenario changes b ecause a
larger amount of data may be transferred. Furthermore, these systems lose
their appeal as candidates for SaaS, or cloud -based implementation, the
more closely they are connected to tangible assets like manufacturing
facilities and warehouse invent ory.
6.4 SUMMARY
In this module we have discussed about the various software architecture
associated with cloud computing technologies. We have learnt Dev 2.0
platform fundamentals such as TCS instant apps on Amazon cloud, along munotes.in

Page 94

Cloud Computing –II
94 with its advantages, disad vantages and its applicability. Followed by
enterprise software: ERP, SCM, CRM. Here we have discussed about the
anatomy of a large enterprise, the products and orders, its execution,
billing and accounting process.
6.5 LIST OF REFERENCES
1. Enterprise Cloud Computing Technology, Architecture, Applications,
Gautam Shroff, Cambridge University Press, 2010.
2. Mastering In Cloud Computing, Rajkumar Buyya, Christian Vecchiola
And ThamariSelvi S, Tata Mcgraw -Hill Education, 2013.
3. Architecting the Cloud: Design Decis ions for Cloud Computing
Service Models (SaaS, PaaS, and IaaS), Michael J. Kavis, Wiley CIO,
2014.
4. Cloud Computing: SaaS, PaaS, IaaS, Virtualization, Business Models,
Mobile, Security and More, Kris Jamsa, Jones & Bartlett Learning,
2013.
6.6 UNIT END EXER CISES
1. Discuss on Dev 2.0 platforms.
2. Write a note on SALESFORCE.COM’s FORCE.COM platform.
3. Explain TCS instant apps on Amazon cloud.
4. State the advantages and disadvantages of DEV 2.0.
5. Illustrate the applicability of DEV 2.0.
6. Write a note on enterprise softw are.
7. What do you mean by anatomy of a large enterprise?]Explain the
products, orders and execution process of an enterprise software.
8. Write a note on billing and accounting process.
9. Explain the concept of enterprise process, build Vs buy and SAAS.

munotes.in

Page 95

95 7
SOFTWARE ARCHITECTURE -II
Unit Structure :
7.0 Objectives
7.1 Introduction
7.2 Custom enterprise applications and Dev 2.0
7.2.1 Software architecture for enterprise component
7.2.2 User interface patterns and basic transactions
7.2.2.1 Layered MVC an d the AJAX Paradigm
7.2.2.2 Common UI patterns
7.2.2.3 Formal models and frameworks
7.2.3 Business logic and rule -based computing
7.2.3.1 What does business logic do?
7.2.3.2 Rule -based computing
7.2.3.3 Modelling business logic using MapReduce
7.2.4 Secur ity, Error handling, Transactions and Workflow
7.2.4.1 Application Security
7.2.4.2 Error handling
7.2.4.3 Transaction management
7.3 Cloud applications
7.3.1 What are cloud based applications?
7.3.2 Examples of cloud applications
7.3.3 Benefits of cloud apps
7.3.4 Types of cloud servers
7.4 Summary
7.5 List of References
7.6 Unit End Exercises
munotes.in

Page 96

Cloud Computing –II
96 7.0 OBJECTIVES
The objective of the section is:
 To get acquaint with the fundamentals of software architecture
 To get familiar with the operating of custom ent erprise applications
and Dev 2.0
 To understand the concepts and terms associated with cloud
applications
7.1 INTRODUCTION
We looked at the various information requirements of a typical business.
We observed that these requirements can, be put into place em ploying
goods or SaaS services. While custom application development may be
necessary for some commercialframework or realm distinct capabilities.
7.2 CUSTOM ENTERPRISE APPLICATIONS AND
DEV 2.0
Here, we will examine the construction of such enterpri se systems,
whether they are created as part of a bundled good or a specially designed
application. We will look at ideas needed to create contemporary apps
during this process. We have discussed the developing Dev 2.0 paradigm,
which makes it possible to create applications largely without the use of
bespoke code. Understanding the internal architecture of Dev 2.0
platforms will also be made possible by our examination practises.
7.2.1 Software architecture for enterprise component
It is possible to think of enterprise applications as being made up of
tasksentities, in charge of group of business activities that have approach
to a common features of information objects.
Technically, each component of an application can be thought of as either
a standalone a pplication or as a subsetof a big corporate task, such as
CRM. The following tasks are carried out by the application (component),
which is crucial for the discussion that follows: It keeps certain data,
facilitates clientreciprocity with the data using di splay level, & does some
business logic calculations while doing so.This type of layered
construction is seen in Figure 1. These architectural layers match the
conventional "model -view -controller" (or MVC) architecture, in which a
control layer divides the user interface from the information model. munotes.in

Page 97

Software Architecture -II
97

Figure 1: Architecture layers
Consider the following architecture as that of Apache Tomcat as an
illustration of how the levelled framework of Fig 1 is deployed in practise:
A portalaccessed to server through H TTP is referred to as the "client." The
application installed on a framework like EC2connected to the browser of
a private network within an organisation. JSPs that gets executed both
make up the presentation layer.Information access layer provides rationa l
level with access to information storage. The controller layer is in charge
of implementing access control, also known as application -level security,
controlling. The complete programme is also accessible to varied
frameworks like web or mashups, in addi tion to these levels.
7.2.2 User interface patterns and basic transactions
7.2.2.1 Layered MVC and the AJAX Paradigm
Informationaccessed and computational screensare used by users to view,
enter, and modify information stored in the data store in order to perform
tasks. Let's think about the actions that are performed duringaddition of a
new entity. We assure a multi -leveldesign, such that shown in Fig 1built
on browsers: The client enters information after accessing a URL that
produces an HTML page. This i nformation must be verified, for instance
to make sure that numerical data is entered into numerical fields.

Figure 2: An order entry form munotes.in

Page 98

Cloud Computing –II
98 Typically, these validations are carried out using JavaScript functions. The
information is then transmitted to the presentation layer server. The user
level, or the informationlaunched into or changed is managed by server -
side presentation layer before delivered to the other levels. It’s effective to
keep part of input information than repeatedly exchanging because th e
HTTP protocol is fundamentally memoryless and asynchronous. We point
out, however, that in a standard client -server architecture, state might be
preserved within the "fat -client" programme itself, negating the
requirement for such server -side functionali ty. As a result, the server -side
presentation layer is once more rendered partly unnecessary.
Let's now examine the controller layer's motive: Think about ainformation
capturing operation once more that requires numerous "pages." How does
the user move aro und the pages? Such are developed by the user
application in a client -server design and the AJAX paradigm. Due to this
architectural limitation, complex controller frameworks like "Struts" and
"Spring" were created.
Let's look at the layered web design. Ma ny of tasks required by MVC are
carried out by the controller layer, which is described above. Layered web
architecture falls short of a key requirement of classic MVC, which is that
the controller must immediately reflect any changes to model information
in the "view" or presentation layer whenever the model's value changes:
When a client approachentity, its most recent record is obviously
displayed. Theyfrequently make connections in the AJAX paradigm,
making a genuine MVC architecture much simpler to imp lement.
Let's now compare the controller layer to the conventional layered to see
how it differs: It is frequently possible to navigate from one page to
another using AJAX without requesting a new HTML page while still in
the same logical database transact ion. As a result, similar to the classic
client -server design, part of the controller's navigational functions is
moved to the client side. The server side continues to be used for the
controller's remaining features, such as interacting with other layers.
Figure 3 depicts the updated layered architecture in the AJAX paradigm.

Figure 3: MVC layering with AJAX
munotes.in

Page 99

Software Architecture -II
99 7.2.2.2 Common UI patterns
We have seen how the user interface of a browser interacts with a server.
Concentrating on the features that such an int erface should have is equally
crucial. There are numerous data model patterns that businesses use
frequently. These can serve as a starting point for outlining the data that a
business application keeps track of. One of these classes is "invoice -
generating " tasks, in which customerapproach, alter, and look for
information while carrying out a business process like "order
management." In contrast, "decision -support" systems use sophisticated
data visualisations based on many types of analytics.
Recollect the order -entry process from Fig 2. When an
upcomingentityrequires to be written, it must be made sure that latestentity
and any inclined transaction are produced. It is necessary to search for,
select, and link to the order object being produced information as well as
the proper billing and delivery addresses.The complexity of the product
price and shipping costs, as well as other factors, should make it evident
that a straightforwardoperation may contain possibly varied formulations.
As a result, even a stra ightforward order entry transaction may have a
complex user interface that required bespoke creation. The creation of
these user interfaces accounts for a sizable portion of application
programming work.
7.2.2.3 Formal models and frameworks
Let's try to id entify some abstractions that could help us more formally
model the behaviour of the order entry transaction. Let's discuss form
technically as collection of fields that map columns in the application
entities. The "order" form, for instance, consists of t he columns customer
name, order date, and order id, each of which corresponds to a column of
the ORDER database.Customer name & Address fields, which correspond
to columns may also be found on another "customer" form. The "order
item" form has fields like order id, quantity, etc. For each form F, let's
declare the basic page

1. Searching page returns R aa a collection of entries that match any
values put into the fields of F.
2. A result page allows users to choose any of shown entity for
manipulating or removi ng while also displaying the outcome of R.
3. A user may alter any F fields for a specific record on an Edit page and
submit their changes to the server.
4. A Create page enables the creation of a new record
Fig 4 shows an illustration of the Search & Result "p ages", which is a
"search orders" screen: Here, despite having the same form, or "order,"
two "pages," namely Search and Result, are combined onto one screen. munotes.in

Page 100

Cloud Computing –II
100 Generally speaking, a "page" of a form may contain other "pages".Keep in
mind that the common fie ld order id connects the two pages from different
forms ensuring it match the specific input.

Figure 4: Search orders form
We now demonstrate how the formal definition of the straightforward
order entry screens shown in Figures 2 and 4 may be achieved.

Here, each line is produced using an abstraction -based formal grammar's
production rule. The type of each token is implicitly defined. With the
help of such a formalism, it is possible to specify the kinds of the fields
that the order form contains as wel l as the forms that are either linked to or
contained in it. For instance, the order_item form's Create and Edit pages
both have aoutcomelayout that lists specific transaction that is generated &
updated.Additionally, both include aninterface that may be u sed to launch
the Create page, allowing users to add new order items while generating or
amending an order. The order_id column, which is shared by the order and
order_item forms, as previously indicated, guarantees that the OrderItem
record being generate d or changed in either case relates to the right Order
record. munotes.in

Page 101

Software Architecture -II
101 Such technical architectural frameworks are frequently created and
maintained in the sector. It is alluring to believe that there might be a
general framework that firms could apply more frequ ently.Unfortunately,
any such structure would quickly become outmoded due to the
development of technology.
7.2.3 Business logic and rule -based computing
7.2.3.1 What does business logic do?
In MVC terminology, the'model' layer includes business logic. As pects
purely connected to presentation are categorically excluded from this
layer. In a tiered MVC design, the controller layer is in charge of
transferring once purified of components. As an illustration, when an
application is sent via a POST, the contro ller layer parses it, extracts the
field values from the REQUEST object that is received, and converts them
& are then accessed and forwarded to the proper functions.In
contemporary web architectures utilising the AJAX paradigm, the server is
required to e xecute requests in addition to sending up new pages and
processing page submissions.
Even within the same application, the business logic layer's functionality
handling can differ greatly. However, similar to user interfaces, there are
some widespread prim itive abstractions that we may also spot in this
situation.
We can list the duties that a business logic method must perform in brief
as follows:
1. Verifications
2. Calculations
3. Agreement monitoring, which involves completing each & every
necessary task with in itscontext.
4. Information manoeuvres and its correspondence mapping
5. Invoking different business logic techniques, such as publishing an
accounting transaction.
6. Complex Search (such as getting several orders and the products
associated with them
7.2.3.2 Ru le-based computing
The abstraction is dependent on fixedrational. In addition to many other
computations, such as assessing risk or rating insurance policies, rules are
particularly helpful for modelling validation rules. When assessed against
facts, a reg ulationmodel comprises rational claimsthat might or might not
turn out to be true.
For instance, the following guidelines could be used to define a "valid"
order: munotes.in

Page 102

Cloud Computing –II
102


Predicates like Order(x) and Customer(z) in the example above evaluate to
T if x, y& z, r espectively, are Order, OrderItem, and Customer objects.
Numerous codes exists for assessing given input as reflected from the
example above.
Rule systems are assessed using rule engines. Two different types of rule
engines exist: Backward -chaining engine s analyse all potential rules that
are necessary to validate if a specific predicate, such as ValidOrder(x), is
true or false. The whole set of available facts are determined by forward -
chaining engines, on the other hand. Applications have typically emplo yed
backward -chaining rule engines.
7.2.3.3 Modelling business logic using MapReduce
We have seen that the business functions of data access and transaction
management may be handled utilising architectural frameworks that
incorporate desired patterns to c onsistently solve these issues. Similar to
what we saw in the previous section, rule -based computing can be used to
abstract validation rules. Computations, data manipulation, calling other
business methods, and complex search are what are left. Some Dev 2 .0
platforms also make an effort to abstract these business logic functions.
For instance, TCS InstantApps leverages the MapReduce cloud
programming paradigm and the Map -reduce -merge model, which extends
MapReduce for relational processing, to create a vis ual formulation called
Logic Map.
A logic map is a graph that contains nodes for create, search, and update
that each represent "pages" of an application form. Without a user
interface, these nodes manipulate relational records in the same way as the
corre sponding pages of their connected forms. These nodes can also be
thought of as map nodes in the MapReduce context; they read and
manipulate database tables while producing new pairs of key/value pairs
(relational records) as input. Additionally, there are "compute nodes"
called merge and reduction that execute calculations on data received after
merging (for merge) or aggregating (for reduce) records flowing along the
edges of a logic map.
The logic map in Figure 5, for instance, executes a billing operatio n for
one or more customer entries. As instances of a customer form C with the
fields cid,...> (cid serving as the primary key of the customer database), munotes.in

Page 103

Software Architecture -II
103 customer records flow into the logic map. This collection of "form
instances" is directed toward a sea rch node connected to the order form O.
The search node gets all orders (from the order table) that were placed by
any of the customers flowing into the node because the attribute cid is
shared by the order and customer forms. Keep in mind that a set of
instances of the order form are now the output of the Search Orders
node.A second search node on a product form P is used in parallel to
retrieve all of the items' current prices. Then, using a temporary
intermediate "valued order" form VO, these parallel re cord sets are linked
together using a merge node. The product prices and quantities ordered are
utilised to calculate the value of each order during this merging operation,
and records from the two incoming sets are linked together using the
common attribu te pid. Instances of the VO form finally flow into a reduce
node on the customer form C, where valued orders are summed to
determine the total billing for each client.

Figure 5: Logic map
Decision nodes comparable to compute nodes are also permitted by l ogic
maps, however they filter records in the flow rather than manipulating
them. Loops can be created using decision nodes, allowing for the
modelling of a reasonably broad class of business logic functions. By
utilising the MapReduce paradigm, it is poss ible to efficiently use cloud
data stores even though they do not explicitly enable joins because the join
capability is included into the merge abstraction. This allows for the
parallel execution of logic maps in cloud environments.
7.2.4 Security, Error handling, Transactions and Workflow
Technical features of an enterprise application that are not immediately
related to functional components also require architectural support. The
elements of the technical architecture shown in Figure 6 carry out the
following tasks:
1. Application security must be used to restrict access to programme
functionality so that only authenticated users can access the
application and use the functions for which they have been granted
permission.
2. For properly handling various clas ses of problems, controlling error
messages, and informing users about issues, uniform error handling is
crucial.
3. In addition to preserving data integrity via underlying database
transactions, transaction management is required to guarantee that
each logic al user interaction is atomic in terms of concurrency control. munotes.in

Page 104

Cloud Computing –II
104 4. Workflow management is required to handle the work that moves from
user to user to complete a business process.

Figure 6: Technical architecture layers
7.2.4.1 Application Security
Whether a client -server architecture or a web -based design, secure user
authentication in a distributed context is a complex topic. The Kerberos
protocol was created to provide authentication across an unsafe network
while still protecting users against replay attac ks and eavesdropping.A
user's password is never sent over the network using this protocol, not
even when it is encrypted or hashed. Instead, a trusted server creates short -
lived tickets and session keys that are used by clients and servers to
connect and t o authenticate one another.Currently, Kerberos variants are
implemented as a fundamental component of distributed security at the
operating system level, in both Windows and Linux, and in Windows
Active Directory and LDAP -based identity management servers.
Each HTTP request sent once a user has successfully logged in comprises
a "user context" that contains the security tokens or tickets needed by the
selected authentication method. Every application service must validate
the user context on the server befo re processing it; typically, the controller
layer ensures this as it processes each request made to the server.
The web application server's underlying features or the application
architecture can be used to provide function access control. Access control
lists (ACLs), which are frequently dependent on the URL or server -side
code packages (such as a Java.jar file) being requested, are typically used
by web application servers to control access to application resources.
However, the fine -grained level of acc ess control required by the majority
of enterprise applications is frequently insufficiently supported by this.
Furthermore, these functionalities cannot be easily used by rich internet
interfaces (RIA), which heavily rely on JavaScript. As a result, funct ion
access control is typically implemented at the application level along with
data access control.
A straightforward data model for function access control is shown in
Figure 7. Through the User Role class, Users have Roles, which are
connected to numero us other Users. Through the Role Function class,
roles have access to functions. Each function is either a "page" of a
specific "form" (such as Create, Edit, Search, etc.) or a menu item. During
login, the application security layer consults the function a ccess control
model and adds the user's list of available functions to the user context. In
order to restrict user access to only those forms and menus allowed by the munotes.in

Page 105

Software Architecture -II
105 function access control model, security must then be checked in both the
presentation lay er and controller layer.Due to the fact that it is
implemented across many architecture layers, application security is an
example of a "cross -cutting" concern.

Figure 7: Function access control
7.2.4.2 Error handling
Users will encounter a wide variety of faults while conducting commercial
transactions, and they must be made aware of these. These include
business failures as well as technical errors that may result from
programme defects, such as when validation tests fail or a notification that
the data being searched for is not available. Additionally, signals like
those that confirm a transaction's success must be communicated together
with pertinent data like the order -id that the system will automatically
produce. An error handling strategy is a set of mechanisms that allow an
application to notify these errors and provide the user the opportunity to
react; an error management framework uniformly applies this strategy to
all user interactions in an application.
An error -handling framework's implementa tion specifics are directly
related to the technological elements and application architectural design
principles. Error handling is thus also a cross -cutting issue, even more so
than application security: All architecture levels between the point where
the issue occurs and the presentation layer are involved when an error or
combination of errors needs to be communicated to the user.
7.2.4.3 Transaction management
From the standpoint of concurrency management, every user interaction in
a transaction -proces sing application that involves adding, deleting, or
altering data must be atomic. It's crucial to understand the distinction
between an atomic database transaction implemented by the controller
layer and an atomic user interaction. The controller compiles all data that a
user might have sent over the course of potentially several HTTP requests.
In order to ensure database integrity, it then sends each of these to the
business logic layer to be carried out as a single database transaction.
However, in a scen ario with multiple users, this might not be sufficient to
guarantee concurrency control: munotes.in

Page 106

Cloud Computing –II
106 Consider a scenario where two users use an Edit form to access a database
record at the same time to reserve a certain resource (like a conference
room).The independen t user sessions each perform their own updates
before sending the changes to the server via HTTP POST requests. To
prevent lengthy transactions, timeouts, and deadlocks, reading a database
record is purposefully kept outside the scope of a database transac tion in
the layered MVC design. So far as sending a distinct database transaction
to the server is concerned, both of these user activities seem to be
successful.
Web -based architectures most frequently employ 'optimistic' concurrency
control using version numbers. Every table has an extra field called ver
that gets bigger with every update. The version number of a record is also
read when it is read. The structure of the update statement as it is carried
out by the data access layer is as follows:

In suc h circumstances, the version number mismatch will result in no data
being discovered (an error), which can be communicated to the user as a
"transaction failed" notice. This happens when the version number has
changed between the time a record is read and the attempt to publish it. As
a result, only one user —the one who acts first —will be successful in
making a reservation for the conference room.
Since each architecture tier needs to keep track of the version number,
which is also conveyed across HTTP requ ests, even though the actual
check is only at the data access layer, optimistic concurrency control,
often known as "soft locking," is a cross -cutting issue.
7.3 CLOUD APPLICATIONS
An application that performs the same functions as a native application but
runs in the cloud and is accessed by web browsers and APIs is referred to
as a cloud application. Typically, when we refer to the cloud, we are
referring to public cloud Infrastructure -as-a-Service platforms; however,
IaaS entered the cloud market much la ter than Software -as-a-Service, with
cloud applications coming first.
7.3.1 What are cloud based applications?
Service -based software Since the late 1990s, cloud apps have existed,
evolving from more basic web programmes that made use of Flash and
Java to offer rudimentary "desktop -like" capabilities available through a
web browser.
"Cloud application" is a vague term. Providing functionality through a
network where computation and storage take place on servers in data
centers is the subject of a definitio n that everyone can agree on. The cloud,
in the broadest definition, is anything that takes place online as opposed to
locally. However, a cloud computing application is typically used in a munotes.in

Page 107

Software Architecture -II
107 more limited sense: it involves infrastructure and applications th at are used
and maintained through the internet, has a web -based user interface, and
frequently — but not always — involves virtualization. This definition
covers both cloud apps that operate in a remote data center and cloud
infrastructure platforms that offer virtual servers, networks, and other
infrastructure.
7.3.2 Examples of cloud applications
A prime example of a cloud application is Google Docs or Office 365. All
you need for Google Docs or Office 365 is a computer that can run a web
browser and an internet connection. Remote servers provide the user
interface and all of the functionality, including data storage. Numerous
distinct cloud apps can be hosted by your company using cloud
application servers.
Cloud applications Vs Native applications
Addit ionally, Google Docs offers a helpful point of contrast between
contemporary cloud applications and the more antiquated native
application paradigm. Bandwidth was limited in the early days of the
internet. Delivering feature -rich applications via the inter net while
maintaining a positive user experience was unattainable.
On local computers, programmes like Microsoft's Office were downloaded
once or purchased on DVDs. The local machine served as the sole
processing and storage facility.
The local application model has some advantages, but in an era with
plentiful bandwidth and a web platform with significantly more
functionality than ever before, many developers opt to design for the cloud
first. Even businesses like Microsoft that gained fame for their deskt op
applications are moving toward the cloud for the distribution of
applications.
More and more companies are utilizing the cloud and cloud software as
access to cloud services and data center IT infrastructure increases for
service providers.
Cloud applic ations Vs Web applications
Nearly as long as the web, web applications have existed. Early web
browsers incorporated JavaScript so that programmers could add
capabilities that went beyond static pages. You'll remember Java -applet
and Flash games and applic ations if you've been using the internet as long
as I have.
What distinguishes cloud applications from web applications then? First,
and rather counter -intuitively, the majority of contemporary cloud
applications leverage web -native APIs and technologies. You are not need
to download a browser plugin in order to access a cloud app because it
makes use of browser functionality. munotes.in

Page 108

Cloud Computing –II
108 Second, while cloud applications compete with native programmes despite
frequently being simpler and providing a more intuitive user experience,
online apps frequently have less capabilities than desktop applications.
7.4 SUMMARY
All of the architectural concerns covered in this section are "cross -cutting
concerns," as we have stressed. Such cross -cutting issues can be
introduced and m anaged in complicated systems using aspect -oriented
programming (AOP), which makes it easier to change the implementation
techniques for such concerns even after an application architecture has
been implemented. AOP, for instance, makes it possible to effe ctively
change cross -cutting functionality for the platform as a whole in Dev 2.0
designs.
7.5 LIST OF REFERENCES
1. Enterprise Cloud Computing Technology, Architecture, Applications,
Gautam Shroff, Cambridge University Press, 2010.
2. Mastering In Cloud Comput ing, Rajkumar Buyya, Christian Vecchiola
And ThamariSelvi S, Tata Mcgraw -Hill Education, 2013.
3. Architecting the Cloud: Design Decisions for Cloud Computing
Service Models (SaaS, PaaS, and IaaS), Michael J. Kavis, Wiley CIO,
2014.
4. Cloud Computing: SaaS, Paa S, IaaS, Virtualization, Business Models,
Mobile, Security and More, Kris Jamsa, Jones & Bartlett Learning,
2013.
7.6 UNIT END EXERCISES
1. Explain the software architecture for enterprise component.
2. Write a note on user interface patterns and basic transact ions.
3. Discuss on layered MVC and the AJAX paradigm.
4. Explain common UI patterns.
5. Explain formal models and frameworks.
6. What do you mean by business logic and rule -based computing?
7. What does business logic do?
8. Write a note on rule -based computing.
9. Explain th e concept of modelling business logic using MapReduce.
10. Elaborate the following concepts: Security, error handling,
transactions and workflows munotes.in

Page 109

Software Architecture -II
109 11. Explain the concept of application security.
12. What do you mean by error handling?
13. Write a note on transaction manag ement.
14. What are cloud -based applications?
15. State the examples and benefits of cloud apps.
16. Explain the types of cloud servers.



munotes.in