Cloud Based Resource Scheduling and Load Balancing

Abstract

ABSTRACT newline newline Cloud computing affords a number of challenges regarding the control of obligations and resources, together with cost constraints and of entirety time necessities. These challenges are even greater for a hybrid cloud computing environment that combines noticeably less expensive however low overall performance private cloud offerings with rather excessive price and excessive performance public cloud services. This thesis provides a solution which mitigates the problem of load balancing in the Cloud. It proposes a new methodology with the objective of improving data availability to enhance the load balancing between the Cloud sites. In summary, this research innovation implicates the design of optimized load balancing algorithms that consider workflow applications. As the first part of the work, Augmented Fish School Search based Stochastic Hill Climbing (AFSS-SHC) for load balancing is proposed. The AFSS-SHC algorithm has also been tested for execution time for Particle Swarm Optimization (PSO), Parallel Particle Swarm Optimization (PPSO), Particle Swarm Optimization - Hill Climbing (PSO-HC), and Augmented Fish School Search Based Stochastic Hill Climbing(AFSS-SHC) for NASA iPSC files and CEA-Curie Files. It can be seen that the performance of AFSS-SHC is better compared to other algorithms. The second part of the work proposes Deadline Constrained Based Resource Allocation in Cloud Environment (DCRA) algorithm for Deadline Management. This work extends by proposing a scheduling based request handler mechanism that optimally manages the requests. Finally the third part of the work is proposes multi parameter optimization for load balancing with effective task scheduling and resource sharing is proposed. A meta-heuristic practice named as Multi-Objective Scheduling based on Particle Swarm Optimization(MOSPSO) that limits the execution cost of the work process while complying with the time constraint in cloud computing condition newline