Dynamic Load Balancing Algorithms in Grid Environment

Abstract

Computer intensive applications require large computational power. These applications can be solved with the single standalone machine, which is very time consuming. Another solution is to use supercomputer but it is not cost effective. The best solution is to use the idle CPU cycles of other nodes which is provided by computational grid. Computational grid can facilitate a user to solve large scale problems by spreading a single large computation across multiple machines irrespective pf physical location. This is the main concept for the grid environment.

The main aim of this project is to distribute the computational load among the available grid nodes. Alchemi toolkit is used for establishment of grid environment. Alchemi Toolkit supports scheduling algorithm which selects the highest priority thread among the pool of threads and dedicate it to the available executor. The priority to the thread is assigned manually.

The scheduling algorithm has been developed which concentrates around the available CPU power. So that scheduling of the thread will be based on the available CPU power so the average waiting time for the thread will be reduced.