Christ University Institutional Repository

DESIGN AND DEVELOPMENT OF LOAD BALANCING ALGORITHM FOR ENHANCING CLOUD COMPUTING PERFORMANCE

SHONEY , SEBASTIAN (2019) DESIGN AND DEVELOPMENT OF LOAD BALANCING ALGORITHM FOR ENHANCING CLOUD COMPUTING PERFORMANCE. PhD thesis, Christ University.

[img]
Preview
PDF (Title) - Published Version
144Kb
[img]
Preview
PDF (Declaration) - Published Version
202Kb
[img]
Preview
PDF (Certificate) - Published Version
541Kb
[img]
Preview
PDF (Acknowledgement) - Published Version
229Kb
[img]
Preview
PDF (Abstract) - Published Version
229Kb
[img]
Preview
PDF (Contents) - Published Version
287Kb
[img]
Preview
PDF (List of figures) - Published Version
224Kb
[img]
Preview
PDF (List of tables) - Published Version
224Kb
[img]
Preview
PDF (Abbreviations) - Published Version
230Kb
[img]PDF (Chapter 1) - Published Version
Restricted to Registered users only

346Kb
[img]PDF (Chapter 2) - Published Version
Restricted to Registered users only

885Kb
[img]PDF (Chapter 3) - Published Version
Restricted to Registered users only

626Kb
[img]PDF (Chapter 4) - Published Version
Restricted to Registered users only

786Kb
[img]PDF (Chapter 5) - Published Version
Restricted to Registered users only

576Kb
[img]PDF (Chapter 6) - Published Version
Restricted to Registered users only

336Kb
[img]
Preview
PDF (Bibliography) - Published Version
423Kb
[img]PDF (Appendix) - Published Version
Restricted to Registered users only

305Kb

Abstract

Software Applications have taken a leadership position in the field of Information Technology to reduce the human workload. In the case of distributed applications, the scalability of the application is a matter of concern in the present dynamic scenario. The fast developments in computing resources have reduced the cost of hardware and increased the processing capability of the system remarkably. Still, hosting a distributed application in a higher end system is not recommended due to many reasons. Firstly, when there is a massive demand in the usage of the application which is beyond the limit of the system, there is no way to scale it. The second reason is that when the system usage of the application is minimal, the entire infrastructure dedicated to the targeted application will remain idle. Due to the wide acceptability of the industry on cloud computing, the variety of applications are designed to target the cloud platform which is one of the challenges for efficient load balancing in the cloud environment. A fair distribution of workload among the available resources is mandatory to improve the efficiency of the cloud platform. To share the workload, a useful load balancing strategy, as well as a timely invocation of the plan, is essential. Invocation of the approach known as triggering policy can be different in centralised and distributed scenarios. Since cloud applications are running in a distributed situation, through this research work, the researcher puts forward a complete framework for balancing load in different types of the request generated in Infrastructure as a Service (IaaS) platform. As a progressive model, this research work continuously focuses on improving the performance of the load balancer in the IaaS platform. Since the cloud data centres are spread across the globe, a centralised monitoring system to monitor and analyse the resource utilisation in different data centres is an essential requirement to see the load fluctuations in different clusters. Even though many open source cloud computing tools are available today with inbuilt monitoring support, a platform-independent monitoring tool will help the research community and the cloud service providers to see the load distribution in different cloud clusters in a realtime environment. In this part of research, Ganglia, an open source tool for monitoring the performance of Grid computing, extended to the cloud environment. Cloud partition model is a hierarchical model widely used for representing a cloud architecture for the public cloud. We performed a detailed study of this model and identified a few significant limitations that restrict the performance of the cloud architecture. The research work proposed a new model for the IaaS platform named cloud partition model with nearest proximity allocation and tested the new model in different scenarios, and the result showed a significant improvement in the performance of the cloud systems. Bringing automation in the load balancing decision process can significantly improve the response time in a significant manner. We introduced a few dedicated and well-defined intelligent agents in the proposed cloud partition model to automate the process of load balancing. All the agents are event-based software agents to handle the tasks. These dedicated agents could bring down the traffic flow happening in the cloud system thereby improving the performance of the cloud systems. The application of queuing theory, to reduce the waiting time in different connected queues, and the implementation of Bees algorithm to reduce the complexity in the search operation enhanced the result of the proposed load balancing algorithm further. The proposed load balancing framework for IaaS platform benefits the non-commercial researchers and non-commercial cloud providers so as to set up an effective load balancing strategy in a cloud architecture.

Item Type:Thesis (PhD)
Subjects:Thesis
Thesis > Ph.D
Thesis > Ph.D > Computer Science
ID Code:7857
Deposited By:Shaiju M C
Deposited On:24 May 2019 11:48
Last Modified:24 May 2019 11:48

Repository Staff Only: item control page