CHAPTER 1

INTRODUCTION

1.1 INTRODUCTION

Cloud computing is an emerging technique for business establishments and

research centre, in the very recent couple of years. It is for the most part shows the

virtualized computing resources with scheduler and the resources in the logical

integration, to manage data center assets, virtualization and client submitted to the

mission requirements for greatest use rate for the client to give administrations

[ZHU12]. Cloud computing is wherever, for example, IT magazines, sites, radios or

TV channels, cloud will get our attention. The present most prominent long range

interpersonal communication, email, archive sharing and network based gaming

destinations, are facilitated on a cloud.

The larger purpose of Microsoft developmental is chipping away at cloud items.

Indeed, even the U.S government expects to introduce cloud-based systems as the

default option for federal agencies of 2012 [JAN13]. Cloud computing is a growing

innovation in appropriated registering which encourages pay per display according to

the client’s request and necessity. Cloud comprises of a gathering of virtual machine

which incorporates both computational and storage facility. The main aim is to give

effective access to remote and geographically distributed resources.

Cloud computing is an innovative progression which helps to design computing

systems, create applications, and utilize existing services to build programs. It depends

on the idea of dynamic provisioning, which is connected to administrations, as well as

1
to process capability, storage, networking and information technology framework.

Today, anybody with a credit card can subscribe to Cloud benefits, and convey and

design servers for an application in hours, develop and contract the foundation serving

its application as indicated by the request, and pay just for the time these resources have

been used. Cloud computing is still under its improvement to organize and has many

issues and difficulties in deciding the compelling execution.

Cloud infrastructure is quickly adaptable, available and furthermore, it is

compensation as we utilize show. Because of the versatile idea of the cloud, the resource

provisioning is possible powerfully as indicated by the essentials of the work process

application. The application of a computation is needed for processing and handling the

huge information with additional warehousing facilities for the resources to be shared

among users.

The advance development in virtualization technologies and cloud computing

servers, the best means for distributing computing resources for existing resource pools

is based on demand and scientific computing. This chapter offers a brief overview of

the Cloud Computing concepts, types and services. The chapter also discusses about

the workflow and scheduling algorithm into developments of Cloud computing

environments. The optimization techniques are introduced in cloud computing to

improve the workflow scheduling methods in diverse applications.

1.2 CLOUD COMPUTING

Cloud computing is an Internet-based process, wherein hardware and software

resources are devoted to clients on request. It is a counter-effect and resultant of the

straightforward entry to remote processing destinations given by the Internet. The term

Cloud refers to a Network or Internet. It states that Cloud is something, which is

2
available in a remote area. Cloud can give benefits over system on open systems or on

private systems (i.e., Wide Area Network (WAN), Local Area Network (LAN) or

Virtual Private Network (VPN)). Cloud computing is mainly used in several

applications such as e-mail, web conferencing, customer relationship management.

Cloud computing uses the web to bring to another person's software running on

another person's equipment in another person's datacenter [RAO13] Cloud processing

refers to getting or figuring resources that are commonly processed and formed by a

third-party on a merged premise in one, or generally more, data center locations. It

highlights on-request provisioning and pay-as-you go resource billing, with negligible

forthright speculation. It has given way for conveying practical figuring control over

the Internet, including virtual private systems. From the viewpoint of a sensible cloud

advocate, cloud administrations limit the capital cost of processing, attach a working

cost to real utilize, and decrease staffing costs.

Figure 1.1 Cloud Computing

3
 Through cloud computing, ready to utilize programming is conveyed through

the Internet on the program with no organization, has an application on the Internet,

sets up own remote file storage and database system and more. Cloud computing gives

us methods, whether we can stick to the applications as utilities, over the Internet. It

enables us to build, design, and modify applications on the network. Cloud computing

refers to controlling, designing, and accessing the applications on the network. It

provides online data storage, infrastructure and application as shown in figure 1.1

The cloud service for several multinational companies in which some of the

notable companies are given below

· Google - It is a private cloud that extends services for email access, document

applications, text translations, maps, web analytics of many users.

· Microsoft - It allows the users of business intelligence tools to move into the

cloud and makes the office applications available in a cloud.

· Salesforce.com - It lets the customers run the applications in a cloud on

Force.com and Vmforce.com products which offer the developers to build

customized cloud services.

1.2.1 Components of Cloud Computing

The components of cloud computing are given below and shown in figure 1.2.

Cloud Components

Client Storage
Application

Services Platform Infrastructure

Figure 1.2 Components of Cloud Computing

4
 · Clients - The client involves computer hardware or software for application

delivery and it is especially designed for delivery of cloud services [DHA12].

· Services - A cloud service includes “products, services and solutions that are

rendered and consumed in real-time over the Internet. For example, Web

Services which may be accessed by other cloud computing components and

software

· Applications - A cloud application leverages the Cloud in software architecture,

often eliminating the need to install and run the application on the customer’s

own computer, thus alleviating the burden of software maintenance, ongoing

operation, and support.

· Platform - A cloud platform, such as Platform as a service, the delivery of a

computing platform, and/or solution stack as a service, facilitates deployment

of applications without the cost and complexity of buying and managing the

underlying hardware and software.

· Storage - Cloud storage involves the delivery of data storage as a service,

including database-like services, often billed on a utility computing basis.

· Infrastructure - Cloud infrastructure, such as Infrastructure as a service, is the

delivery of computer infrastructure, typically a platform virtualization

environment, as a service.

1.3 ARCHITECTURE OF CLOUD COMPUTING

Most cloud storage providers generally follow a three-layer architecture as

shown in figure 1.3

· Front End - The front end is in charge of the communication between the clients

and the servers. There are different APIs to access the actual storage. It consists

5
 of interfaces and applications that are required to access the cloud computing

platforms, e.g., Web Browser

· Storage Layer - The storage logic layer handles a variety of features, and is in

charge of certain administrative procedures such as ensuring a high level of

availability and reliability. It is also a form of security parameter. Furthermore,

it acts like a controller for cloud storage.

· Back End - It refers to the cloud itself. It consists of all the resources required

to provide cloud computing services. It comprises of huge data storage, virtual

machines, security mechanism, services, deployment models, servers, etc. The

back-end focuses on the actual implementation of the physical storage of data

with protocols such as the GFS (Google File System). It involves the use of

various ways to increase storage efficiency and in a way to drive the

infrastructure costs down.

Client Infrastructure Front End

Internet

Application

Service
Management

Cloud Runtime
Security

Back End

Storage

Infrastructure

Figure 1.3 Cloud Computing Architecture

6
1.3.1 Characteristics of Cloud Computing

Cloud computing allows for the sharing and scalable deployment of services, as

needed, from virtually any location, and for which the customer can be billed based on

actual usage. Cloud computing has a variety of characteristics which are

· On-Demand Self-Service - A service consumer can automatically make use of

the computing capabilities, such as server processing time and network storage

without requiring human interaction with each service provider.

· Broad Network Access - Cloud capabilities are available over the network and

accessed through various platforms (e.g., mobile phones, laptops, and tablets)

[GAM13].

· Measured Service - Cloud systems automatically control and optimize

resources by leveraging a metering capability in which resource usage can be

monitored, controlled, and reported, providing transparency for both the

provider and the consumer of the utilized service. The advantage here is that

you are paying for just what you are habituating.

· Shared Infrastructure - Uses a virtualized software model, enabling the sharing

of physical services, storage, and networking capabilities. The cloud

infrastructure, regardless of the deployment model, seeks to make the most of

the available infrastructure across a number of users.

· Dynamic Provisioning - Allows for the planning of services based on current

demand requirements. This is performed automatically using software

automation, enabling the expansion and contraction of service capability, as

needed. This dynamic scaling needs to be done while maintaining high levels

of reliability and security.

7
 · Network Access - Needs to be accessed across the internet from a wide range of

devices such as PCs, laptops, and mobile devices, using standards-based APIs

(for example, ones based on HTTP). Deployments of services in the cloud

include everything from using business applications to the latest application on

the newest smartphones.

· Managed Metering - Uses metering for managing and optimizing the service,

and to provide reporting and billing information. In this way, consumers are

billed for services according to the use they have made actually during the

billing period.

1.3.2 Benefits of Cloud Computing

The cloud makes it feasible for one to get to one’s data from anywhere,

whenever it is required. While a conventional PC setup expects to be in an

indistinguishable area of data storage device, the cloud makes away that stride. The

cloud expels the requirement for one to be in an indistinguishable physical location of

the hardware that stores the data. The cloud provider can both possess and house the

hardware and software that are important to maintain one’s home or business

applications. Little organizations can store their data in the cloud, evacuating the cost

of purchasing and putting away memory gadgets. Some of the benefits of cloud

computing are shown in figure 1.4.

8
 Figure 1.4 Benefits of Cloud Computing

The following are some of the possible benefits of those who offer cloud computing-

based services and applications [VOO11] are

· Cost Savings - Companies can reduce their capital expenditures and use

operational expenditures for increasing their computing capabilities. This is a

lower barrier to enter and also requires fewer in-house IT system resources to

provide system support.

· Scalability/Flexibility - Companies can start with a small deployment and grow

to a large deployment fairly rapidly, and then scale back, if necessary. Also, the

flexibility of cloud computing allows companies to use extra resources at peak

times, enabling them to satisfy consumer demands.

· Reliability - Services using multiple redundant sites can support business

continuity and disaster recovery.

9
 · Maintenance - Cloud service providers manage the system maintenance, and

access is through APIs that do not require application installations onto PCs, so

further reducing maintenance requirements.

· Location independent, so long as there is access to the Internet

· Increased security at a much lesser cost when compared to traditional stand-

alone applications due to centralization of data & increased security focused

resources

· Easy to maintain as they don’t have to be installed on each user’s computer

1.4 CONCEPTS OF CLOUD COMPUTING

There are services and models working to make cloud computing possible and

reachable to end users. Subsequent are the working models for cloud computing:

· Service Models

· Deployment Models

· Virtualization

1.4.1 Service Models

The cloud services can be sorted into software services and infrastructure or

hardware services. Regarding development, programming in the cloud is substantially

more advanced than the equipment in the cloud. Once a cloud is set up, how its

distributed computing administrations are sent regarding plans of action can vary

contingent upon prerequisites are shown in figure 1.5. The primary service models are

defined as

10
 Figure 1.5 Service Model of Cloud Computing

· Software as a Service - A SaaS provider gives subscribers access to the

resources and applications. SaaS makes it unnecessary to have a physical

duplicate of software to introduce the devices. SaaS likewise makes it less

demanding to have a similar programming on the greater part of the device on

the double by getting into it on the cloud [HUT11]. In this model, an entire

application is provided to the client, as an administrator on request. A single

instance of the service keeps running on the cloud and various end clients are

adjusted. On the customers‟ side, there is no requirement for forthright interest

on servers or programming licenses, while for the supplier, the expenses are

brought down, since just a single application should be facilitated and kept.

Today,SaaS is offered by companies such as Google, Salesforce, Microsoft,

Zoho, etc.

· Platform as a Service (Paas) - It is a software or development layer which offers

service, on which the high levels of service can be built and it allows the

customer to build his own applications on cloud infrastructure. The Paas offers

11
 the integration of OS and application servers includes LAMP platform (Linux,

Apache, MySql and PHP), restricted J2EE, Ruby etc. to face the scalability

necessities and management of the applications. Some of the examples of PaaS

are Google’s App Engine, Force.com, etc.

· Infrastructure as a Service (Iaas) - IaaS delivers basic storage and computing

capabilities,networking equipment, data center space as standardized services

and to manage workloads over the network. The customer builds his own

software on the cloud environment. Some common examples are Amazon,

GoGrid, 3 Tera, etc.

1.4.2 Deployment Models

The cloud computing is deployed on infrastructure which differs based on

requirements and characteristics and that which supports the cloud users and service

requirements. The following four deployment models differ in various ways are shown

in figure 1.6 [RAO13]

Public Cloud

Public cloud or External cloud portrays distributed computing in the customary

standard sense, whereby resources are progressively provisioned on a fine-grained, self-

service premise over the Internet, through web applications/web administrations from

an off-webpage outsider supplier who charges on a fine-grained utility registering

premises. The cloud framework is accessible to the overall population or a vast industry

for gathering, and is possessed by an association offering cloud administrations.

Illustrations: Amazon Elastic-Compute-Cloud, IBM’s Blue-Cloud, Sun Cloud, Google

AppEngine.

12
Community Cloud

A community cloud might be built up where a few associations have

comparable requirements and expect to share foundation in order to read a lot of the

benefits of cloud computing. With the costs spread over less clients than a public cloud

(even more than a sole occupant) this choice is costlier; yet may offer a larger amount

of privacy, security and/or policy compliance. Examples of group cloud incorporate

Google’s “Gov Cloud”.

Figure 1.6 Deployment models of Cloud Computing

Hybrid Cloud

The term “Half breed Cloud” has been used to mean either two separate clouds

consolidated (open, private, inner or outside), or a mix of virtualized cloud server

occurrences utilized together with actual physical hardware. The most right meaning of

the expression “Cross breed Cloud” is the utilization of physical equipment and

virtualized cloud server occurrences together to make a solitary basic administration.

Two clouds that have been consolidated are all the more effective called a “combined

cloud”. A hybrid storage cloud uses a combination of public and private storage clouds.

13
Hybrid storage clouds are regularly valuable for archiving and backup functions,

allowing local data to be replicated to a public cloud.

Private Cloud

A private cloud is a specific model of cloud computing that includes a distinct

and secure cloud based condition in which just the predetermined customer can work.

Similarly to the other cloud models, private clouds give figuring power as an

administrator in a virtualized situation utilizing a hidden pool of physical processing

asset. In any case, under the private cloud, the cloud (the pool of asset) just opens by a

single organization providing that organization with greater control and privacy.

1.4.3 Virtualization

The basic virtualized model consists of cloud users, service models, virtualized

model and its host OS and their hardware. The virtualization is the process of designing

a virtual version of a device or resource, which divides the resource into one or

more execution environments such as a server, storage device, network and operating

system. Devices, applications and human users can interact with the virtual resource.

For instance, hard drive also considered as virtualization as it consists of one drive

and partitions it into two separate hard drives. The term virtualization has became a

slogan because it has capabilities to associate with a number of computing technologies

which are given below

· Storage Virtualization - The process of merging the multiple network that

consists of storage devices into a single storage unit.

· Server Virtualization - It includes the partitioning of a physical server into

smaller virtual servers.

14
 · Level Virtualization - It is a type of server virtualization technology , which

works at the operating system (kernel) layer.

· Network Virtualization - This virtualization uses the network resources by the

logical segmentation of a single physical network.

· Application Virtualization - This virtualization works in the application

process.

1.5 CHALLENGES OF CLOUD COMPUTING

Even the innovative technology of cloud computing leads to concerns about

several challenges. Some common challenges are [HAR10]

Data Protection

Data Security is an essential component that warrants examination. Ventures

are hesitant to purchase an affirmation of business information security from merchants.

They fear losing information to rivalry and the data privacy of purchasers. They fear

losing data to competition and the data confidentiality of consumers. In the current

models, firewalls crosswise over s (claimed by ventures) ensure this delicate

information. In the cloud demonstrate, Service providers are in charge of keeping up

data security and endeavours would need to depend on them.

Data Recovery and Availability

The business applications have some service level agreements to be followed

mandatory. The Operational services are essential in service level agreements and

runtime of applications. It supports some of the process for data recovery and

availability.

15
They are

Ø Appropriate clustering and Failover

Ø Data Replication

Ø System monitoring (Transaction monitoring, logs monitoring and others)

Ø Maintenance (Runtime Governance)

Ø Disaster recovery

Ø Capacity and performance management

Management Capabilities

Regardless of there being various cloud providers, the administration of the

stage and the framework is still in its early stages. Components like Auto-scaling for

instance, are a pivotal prerequisite for some ventures. In that perspective, there should

be an immense potential to enhance the scalability and load balancing features

highlights.

Regulatory and Compliance Restrictions

In European nations, Government controls don't permit client's personal

information and other data to be physically available outside the state or nation. In order

to meet such prerequisites, cloud providers need to set up a or a storage capacity site

only inside the country to agreeing to the controls. Having such a foundation may not

generally be durable and is a major test for cloud providers.

16
1.5.1 Advantages of Cloud Computing

Some of the important advantages of cloud computing are given below

· Lower Space - Most of the cloud computing is web based which allows to run

on the desktop, PC, laptops, tablets and it does not need the memory space or

processing power.

· Improved Performance - Due to less programs and fewer processes which are

loaded into memory, it allows the computers to reboot and run faster in a cloud

environment.

· Reduced Software Costs - Instead of purchasing licensed software applications,

it can be accessed for free.

· Unlimited Data Storage Capacity and Increased Data Reliability - The cloud

computers have the ability to store hundreds of Petabytes of data. The cloud

crashing among the cloud computers does not affect the storage of data.

· Universal Document Access - It allows to access the document in the cloud by

an Internet connection

· Device Independence - It allows for changes to the computers, applications and

documents from the cloud system.

1.5.2 Applications of Cloud Computing

The sharing and management of resources of cloud make the cloud computing

as a prominent way for online computing resources [KUM11]. It is an active component

in the various fields as follows

17
· E-Learning - E-Learning is an emerging attractive environment for students,

faculty members and researchers and allows them to connect to their own

organization and access data from the cloud.

· Enterprise Resource Planning (ERP) - The ERP allows the organization to

improve its business through cloud. The installation of the ERP in the cloud

supports to handle more complex and expensive applications, human resources,

payroll etc.

· Backup and Cloud Computing - The backing up of a website or important data

can be achieved by paying a monthly fee so that the customers can receive

storage and bandwidth to do backup and recovery of information.

· Cloud Computing in Office - Cloud applications are becoming common usage

in the office like DaaS (Desktop as a Service) or SaaS and it supports for

reducing the licensing costs.

· E-Governance - E-Goverance makes the functioning of a government more

scalable and customized to offer the services to its citizens, institutions and to

cooperate with other governments by. It reduces the issues of handling,

installing and upgrading E-Governance applications.

· Aadhaar (Unique Identification of People) - The cloud computing makes

prominent way for issuing the UID (Jan Lokpal Bill) one and only single Id used

instead of Voter ID Card, Address Proof like Electricity Bill, Ration Card etc to

access services such as obtaining a bank account, passport, driving license and

so on.

· Health Care - National Rural Health Mission is a web based application

employed in cloud environment which permits people to access Hospital's

18
 Database for finding the nearby hospital location and to get advice from the

doctors through Audio/video/web conferencing.

· Agriculture - Agriculture and the handicraft industry workers are facing issues

to sell their own products in the market which leads to exploitation of the

farmers.

1.6 WORKFLOW IN CLOUD COMPUTING

With the advancement of the world's driving organizations, cloud computing is

drawing in more consideration for giving an adaptable, on demand computing

framework for various applications [DEE03]. The idea of cloud computing keeps on

spreading broadly, as it has been acknowledged well of late. Cloud computing has

numerous exceptional favourable circumstances which can be used to encourage the

work process execution. Moving work processes to a cloud computing condition

empowers the usage of different cloud administrations to encourage the work process

execution. Common cases of work processes incorporate internet online banking,

insurance claim processing [BAL11].

A workflow is a sequence of connected instructions. The intention of workflow

scheduling is to computerize the systems, particularly which are engaged with the way

toward passing the data and the records between the members of the cloud, keeping up

the imperatives. Workflow scheduling has a Non-deterministic Polynomial (NP) issue,

as it is difficult to create an ideal arrangement inside polynomial time and calculations

or to concentrate on producing inexact or close ideal arrangements [CHA12]. The

multi-target nature of the booking issue in Clouds makes it hard to take, particularly on

account of complex occupations like work processes. Workflow scheduling is one of

the key issues in the administration of workflow execution.

19
 A workflow empowers the organizing of applications in a directed acyclic graph

shape where every nodes speak to the constituent task and edges speak to represent inter

task conditions of the applications .A single workflow for the most part comprises of

an arrangement of undertakings each of which may speak with another assignment in

the work process. A workflow for the most part engaged with computerization of

methodology. Workflow scheduling is a kind of task scheduling booking which

concentrates on mapping and dealing with the execution of reliant assignments on

shared resources that are not straightforwardly under its control. A workflow

management system oversees and executes work processes. It incorporates five

measurements that are time, cost, fidelity, reliability and security [KUM13].

Figure 1.7 Workflow Management System

20
 The significance of the cloud is its application adaptability or versatility. This

nature of cloud encourages changes of asset and qualities at run time. This capacity of

cloud empowers the work process administration frameworks to promptly meet the

nature of administration prerequisites. The Workflow Management Coalition (WFMC)

characterizes workflow as [SIN13], recognizing the interfaces inside the structure

which empowers items to interoperate at an assortment of levels as appears in

figure 1.7. This model is characterized as a workflow management system and the most

critical framework interfaces. The figure 1.8 demonstrates the reference model of the

work process as

· Workflow Engine - Software that gives the run-time condition, keeping in mind the

end goal to make, oversee and execute work process occurrences.

· Process Definition - The portrayal of a work process in a shape which underpins

robotized control.

· Workflow Interoperability - Interfaces to help interoperability between various

workflow systems.

· Invoked Application - Interfaces to help connect with an assortment of IT

applications. Work process Client Application - Interfaces to help communicate

with the UI.

· Administration and Monitoring - Interfaces to give framework checking and metric

capacities to encourage the administration of composite workflow application

conditions.

21
 Process Definition

Interface 1

Interface 4
Administration & Interface 5 Workflow Engine(s)
Wo
Workflow Engine(s)
Monitoring tasks

Interface 2 Interface 3

Workflow Invoked
Client Applications
Application

Figure 1.8 Workflow Reference Model

22
1.6.1 Workflow Scheduling Problem

A workflow application is represented as a Directed Acyclic Graph (DAG) in

which graph nodes undertake the tasks and the edges speak about information

conditions among the tasks with weights on the nodes representing the computation

complexity and weights on the edges representing communication volume. In this way,

workflow scheduling problem is typically considered as an exceptional instance of the

DAG planning issue, which is a Non-deterministic polynomial (NP) - finish issue

[RAH13]. Accordingly, despite the fact that the DAG scheduling problem can be

explained by utilizing thorough inquiry strategies, the complexity of generating the

schedule turns out to be high.

The general finish/completion time of an application is typically called the schedule

length or makespan. In this way, the goal of workflow scheduling techniques is to limit

the makespan of a parallel application by a legitimate portion of the assignments to the

processors/assets and course of action of undertaking execution arrangements. Let us

assume workflow W (T, E) consists of a set of tasks, T= {T1, T2,…Tx… Ty, Tn} and a

set of dependencies among the tasks, E= {< Ta, Tb >, ...,< Tx, Ty >}, where Tx is the

parent task of Ty. R = {R1, R2,… ,Rx, ,Ry,Rm} is the set of available resources in the

computational grid. Therefore, the workflow scheduling problem is the mapping of

workflow tasks with grid resources (T →R) so that the makespan M is minimized.

For the most part, a workflow task is an arrangement of guidelines that can be

executed on a single processing element of a registering asset. In a workflow, an entry

task does not have any parent errand and a leave assignment does not have any child

task. Also, a child task can't be executed until the point that the greater part of its parent

task is finished. In terms of scheduling, the task that has the greater part of its parent

task completed is known as a Ready task.

23
1.7 SCHEDULING IN CLOUD COMPUTING

Resource allocation and scheduling of resources have been a vital viewpoint

that influences the execution of system networking, parallel, distributed computing

and cloud computing. Numerous analysts have proposed different algorithms for

assigning, planning and scaling the assets proficiently in the cloud. Scheduling

processes in the cloud can be summed up into three phases as follows:

· Resource Discovering and Filtering - Datacenter Broker finds the assets

display in the system framework and gathers status data identified with

them.

· Resource Selection - Target resource is chosen in view of specific

parameters of task and resource. This is choosing stage.

· The Task Submission - Task is submitted to the resource selected.

For a case scheduling the tasks/Jobs at the server end is exceptionally

troublesome for a server, on the grounds that the number of jobs asking for is

extensive in number, which required a few assets to execute. The timetable along

these lines that works by the server must be idealistic and sufficient with the goal

that each demand by the client gets reaction in time, and each Task/Job gets

appropriate assets for its execution. In Cloud, there are distinctive servers over the

cloud which takes customer's solicitations and after execution offers react to the

customers. The ideal Workflow Scheduling for such framework is given in figure

1.9

24
 Figure 1.9 Examples for Job Scheduling

A large portion of these servers is based on various Virtual Machines (VM).

Virtual machines take input from clients and after the execution responds

[SIN13]. Thought about taking the servers as VMs and these VMs react to

customer's solicitations.

Steps for Job Scheduling

Step-1 All the incoming tasks will be grouped together as according to their
certain behavior or attribute (like deadline constrained or low cost
requirement).
Step-2 After grouping of task, the tasks will be prioritized. A task can be
prioritize on the basis of its attributes like deadline, execution time etc.
Step-3 The Virtual machine with minimum response time and is capable to
execute selected task will be assigned for the execution of the task.

25
1.7.1 Limitations of Existing Workflow and Scheduling Mechanisms

In general, a workflow scheduling strategy can be either static or dynamic.

Static scheduling maps assets to each individual task before workflow execution; it

bolsters the entire work process execution .However, it relies upon the data of

workflow applications and execution circumstances .On the contrary, dynamic

scheduling settles on resource mapping choice exactly when an endeavour is set up to

execute without requiring any prior application and condition learning.

In dynamic scheduling, the task dependency is managed by the job submission

supervisor which approves task conditions. The DAG task in the line meets the

dependence need and is not isolated from other free vacations. For example, in the

Condor system, DAG checks between the errand reliance and submits only the

assignment arranged to execute into the line administered by Condo, which designs the

occupations in the line in First Come First Service (FCFS) outline. Concerning static

heuristics of DAG getting ready for a heterogeneous space, there are two social

occasions that are more common than others, for example, List booking heuristics and

Gathering heuristics. A summary of booking heuristic keeps up a once-over of all

endeavours of a given DAG as demonstrated by the necessities. It immediately sorts

out the positions of all errands, and then picks best resource (by portraying cost target

work) for the readied task with most lifted need.

The clustering algorithm incorporates a two-stage scheduling approach. The

underlying stride is the clustering phases are errands are accumulated into groups of

virtual processors for particular criteria, for example, remembering the true objective

to cover silly trades. The second one is the mapping stage, wherein all part errand of

26
same bunches is allocated to the same processor. Dominant Sequence Clustering (DSC)

and Clustering and Scheduling System (CASS) also have a place with this gathering.

Wide comparative examinations are performed and they testify that the static

framework can possibly perform close perfect, and can exhibit legitimate with some

certifiable work process applications. The entertainment work also prescribes that static

philosophies still perform better than dynamic ones for data raised work process

applications ,even with off course information about future courses. Nevertheless, it is

astoundingly difficult to gage execution accurately, since the execution condition may

change a significant measure after resource mapping. Late work exhibits that arranging

through resource reservation and execution can ensure the advantage of availability in

the midst of execution and speculatively makes the lattice all the more obvious.

However, their philosophies don't resolve each one of the issues. Others endeavour to

make the static approach more adaptable to change occurring in execution arrange.

Rescheduling is actualized in the Grid Analysis and Display System (GrADS),

where it is normally incited by contract infringement. In any case, the attempts are out

and out driven for iterative applications, empowering the system to perform

rescheduling decisions at each accentuation. The arrangement exchanging approach

incorporates building a gathering of activity graphs beforehand and explores the

techniques for changing the beginning with one individual from the family, then onto

the following when the execution of one development chart flops, however every one

of the plans are done without learning of the future condition change. Another

rescheduling game plan is proposed, which considers rescheduling at a couple,

purposely picked centers in the midst of the execution.

27
1.8 OPTIMIZATION TECHNIQUES

User applications may cause huge information recovery and execution costs

when they are planned considering just the 'execution time'. In addition to optimizing

the execution time, the cost emerging from information exchanges between resources

should be considered. Since the conventional methodologies utilized as a part of

improvement given can't be relevant in distributed computing or present shortcomings

.Current heuristic based calculations are produced and have been turned out to be

appropriate for work process booking. The improvement procedures such as Grey wolf

Optimization and African Buffalo Optimization are utilized based on the calendar and

allotment for distributed computing.

Optimization techniques are the procedures of maximization or minimization of

at least one capacities with any conceivable limitations. It can be grouped in the light

of the kind of imperatives, nature of outline factors, physical structure of the issue, idea

of the conditions included,the deterministic nature of the facts, allowable estimation of

the plan factors, separability of the functions and number of objective functions. The

main components of an optimization problem are discussed as

Objective Function

The procedure of maximization or minimization of the target work is performed

on the at least one amount. The target work confronts the issues which have the single

target capacity or more target works yet the distinctive goals are not good. The factors

that advance one goal might be a long way from ideal for the others. The issue with

multi-targets can be reformulated as single target issues by either framing a weighted

blend of the diverse goals or by regarding a portion of the destinations as requirements.

28
Variables

A set of unknown elements are called variables and it can be continuous, discrete

or Boolean which are used to define the objective function and constraints. The design

variable cannot choose randomly because it has to satisfy the certain functional and

other requirements of variables.

Constraints

The unknowns sets are allowed to analyse the certain values by a set of

constraints which are dependent on certain conditions that should satisfy the conditions

of the optimal design.

1.9 PROBLEM DEFINITION

Several algorithms are discussed, but since some of the problems occurred,

hence while referring the data mining the efficiency, accuracy and the speed should be

considered. In some cases, the problem can be rectified by unique and static datasets,

but in real time datasets it is difficult to find the solution. In order to enhance the

workflow scheduling accuracy there is a need for the introduction of optimization

technique and to find the process failure and to improve learning algorithm efficiency.

The various methods and current platforms, and the challenging problems in dynamic

resource scheduling based on cloud computing require to reduce the user request’s

response time and to balance the workloads, to improve the availability and reliability

of the resource.

29
1.10 MOTIVATION

Scheduling is the most challenging tasks in cloud computing platform. It is

important to implement the scheduling algorithm because of high process efficiency,

fast response time, high throughput and cost effectiveness. It defines the strategies

which are used to allocate the processor. Therefore the scheduling and management of

resources in cloud environment are complex and risky and the demands for

sophisticated tools are also high. There is a need for an algorithm that improves the

availability and reliability for workflow scheduling based on cloud computing. The

cloud organization is designed in such a way to support huge database, in e-business

and e-science real time applications.

1.11 OBJECTIVES

The main objectives of this work are

· To improve the features of optimization algorithm to make it suitable for

clustering data.

· To evaluate a new Clustering algorithm in order to work under parallel

computing.

· To analyse the data that has not been carefully screened and represent the quality

of data.

· To discover structures and user patterns in real time high dimensional data set

by avoiding irrelevant data.

· To create a novel learning algorithm as to identify the unique features

effectively

· To set certain limitations to each resource, so that the resource allocation and

workload can be maintained in an optimal way.

30
1.12 ORGANIZATION OF THE WORK

The thesis is organized as follows

Chapter 1 deals with the introduction of cloud computing and workflow scheduling

techniques.

Chapter 2 reviews the previous work done in the areas of cloud workflow

scheduling techniques classified as Best effort based scheduling and QoS based

scheduling.

Chapter 3 deals with the brief description of the “Research Methodology” of the

proposed approaches.

Chapter 4 discusses the first approach “Implementation of Chronic inflation Based

Power Iterative Clustering Algorithm”.

Chapter 5 describes the second approach, namely, “Workflow Scheduling using

Grey Wolf Optimization in Cloud Computing”.

Chapter 6 deals with scheduling technique, “Dynamic Scheduling Algorithm with

African Buffalo Optimization” to improve the efficiency of scheduling in terms of

time, cost and accuracy

Chapter 7 concludes the proposed approach of workflow scheduling, findings and

the future work of the research.

The works of several researchers are quoted and used as evidence to support the

concepts explained in the thesis. All such evidences used are listed in the Reference

section of the thesis.

31
1.13 SUMMARY

This chapter illustrates the cloud computing concepts, workflow and scheduling

problems and limitations of existing scheduling algorithm. In this research, the

workflow scheduling problem in a cloud computing environment is defined and the

existing workflow scheduling problems are discussed. In order to overcome the

limitations, the optimization techniques are introduced to improve the workflow

scheduling. The problem statement, objective and organization of the thesis are also

briefly summarized. The work done by various researches are studied, analysed and

tabulated as some important literature’s in the next chapter.

32