LOAD BALANCING IN CLOUD COMPUTING

Abstract

The advent of cloud computing in recent years has sparked an interest from different organizations, institutions and users to take advantage of services and applications. Because of serving a very attractive package of services, cloud technology has grasped a huge attention from academia, IT industry and government organizations. Cloud computing promises scalability and on-demand availability of resources. As the number of users on the internet goes on increasing, it becomes difficult to handle the millions of user requests on it. Many a times it happens during the peak-hours that the traffic on the network increases abruptly. In such situations it is commonly observed that the system performance degrades. Though having many good features such issues restrict a user to use the services. Users may not find it that much reliable. Therefore, one of the important issues which need a major consideration of the researchers is load balancing in cloud computing systems. A number of load balancing algorithm are proposed by various researchers, to solve this problem. Two kinds of load balancing algorithms are there one is static and other is dynamic. A cloud computing system is supposed to handle the request dynamically rather than doing it by a static approach which is supposed to be the less efficient approach. Also there are number of parameters like resource utilization, throughput, scalability, flexibility, response time, which are used to validate any load balancing algorithm. Any load balancing developer is supposed to maintain a good of trade-off between these parameters. In this thesis a new approach for load balancing in cloud computing is proposed. This algorithm aims at distributing the equal load on each server in the cloud network. This also improves the resource utilization. With the proposed approach a situation in which only few resources are loaded heavily while others are just sitting idle will never arise. The proposed algorithm is also compared with the existing load balancing techniques.