An Algorithm for Load Balancing in Computational Grid Using Activity-Based and CPU Usage Approach

Abstract

Grid computing is a collection of networked computers which work together to solve large problems. Grid becomes complex if resources are geographically spread over wide area, dynamic, and heterogeneous network. This makes load adjusting and adaptation to internal failure more critical in the event of processing network and determining which job requires resource to be assigned and how the load allocated among different machines to improve performance of grid computing. We introduce Dynamic Load Balancing algorithm which takes the resources to be assigned to the job according to the queue length, CPU usage, and free memory. In this research paper we reduced the execution time of application for completion of particular task. When system performing task in between system fail then machine of load transfer from one machine to another machine. We found that Dynamic Load Balancing algorithm is used for matrix users which aims to execute their program speedily using minimum execution time and minimize the idle time of processors.