Load Balancing and Job Migration in Grid Environment

Abstract

Grid Computing is emerging as wide scale distributed infrastructure that promises to support resource sharing and synchronised problem solving in dynamic and hetrogeneous environment. Grid Computing is becoming a generic platform for high performance and distributed computing due to which it is being adopted in various areas like academic, industry, and research use. Grid Computing has progressed a lot, yet, there are some areas of concern, like resource management, resource scheduling, load balancing and security over which, research is still in progress. Load balancing is a great challenge in dynamic and heterogeneous environment like Grids. Load balancing is a technique to enhance resources, utilizing parallelism, improve throughput, and to cut response time through an appropriate distribution of the applications. The main goal of load balancing is to provide a distributed, low cost, scheme that balances the load across all the processors. An effective and efficient load balancing algorithm is required to balance the load in Grid environment, which is a tedious task due to basic nature of Grid environment. Focus of this thesis is to analyze all the existing load balancing algorithms and to do a comparative study. After considering all pros and cons of all the existing algorithms,a new load balancing algorithm has been proposed based on all positive aspects of existing algorithms.The proposed algorithm has been implemented and tested in real Grid environment establised at Research Lab CSED,TU. Along with this ,a job migration technique for balancing the load in Grid Environment has also been discussed.