Design and Development of Effecient Load Balancing algorithm for Performance improvement in cloud environment

Abstract

v newlineABSTRACT newlineNow-a-days, cloud network is an emerging technology that provides various services to its clients over the Internet. Therefore, number of client requests increases where the server or physical machine (PM) in the datacenter is limited. To handle large number requests, virtualization concept is used where one server is logically divided into number of virtual machine (VM) and tries to handle all incoming user requests. Due to dynamic nature of cloud network, loads are fluctuated with respect to time for allocating a suitable VM. It creates a problem in balancing the load balancing within VM. This problem leads to high cost and minimize the profit of an organization and degrade system performance. Each physical machine consumes electric power for the work. That means if a server has number of workloads, then it consumes more energy. To overcome this problem, a better scheduling algorithm is required that can handle the load among the VMs being in balanced state and execute all the incoming tasks within less execution time and consume less energy on the datacenter. The main objective of the thesis is to develop an intelligent action that handle the extra task by transferring from overloaded VM to the underloaded machine to maintain the load of all VM balanced. It also reduces the task execution time of all incoming task that can be directly affected to minimize the energy consumption of the available server in a datacenter but maximize the utilization of resources. newlineThis thesis proposes three different scheduling techniques to solve the load balancing problem. These scheduling techniques are based on meta-heuristic optimization method such as Particle Swarm Optimization (PSO), Parallel PSO and machine learning method such as Deep Reinforcement Learning (DRL) and Actor-Critic method. newlineFirst scheduling method is named as Load Balancing Using Modified Particle Swarm Optimization (LBMPSO) which is used to decrease the overall makespan time and increase utilization of resources. Second scheduling method is based on newlinevi newlinehybrid scheduling method named as Hybrid Particle Swarm Optimization Actor Critic (HPSOAC) which is used to solve an energy consumption problem in datacenter. Third scheduling method is based on parallel scheduling method named as Deep Reinforcement Learning with Parallel PSO (DRLPPSO) to increase the speedup value that reduce the overall task processing time. newlineFrom the simulation results it is observed that proposed methods shown better result as compare to existing algorithms. These methods help the cloud users to access their own service with a great benefit. newline