RESOURCE BROKER MANAGEMENT BY ADOPTING SELF-ADAPTIVE MULTI-INSTANCE BROKER SCHEDULING IN GRID COMPUTING

Abstract

A grid resource broker seeks to assign the appropriate jobs to the appropriate resources as part of resource management in the multi-grid environment. Multi instances of the broker system provides multiple instances of brokers to simultaneously process jobs between multiple resources in a hierarchical cluster grid environment. In this study, the multi-instance broker is developed using grid resource broker taxonomy properties. The number of broker instances to be used for each processing session is determined by calculating resources, computing power and workload. The Self-Adaptive Multi-Instance Broker Scheduling algorithm SAMiB was tested against iHLBA algorithm through four types of scenarios containing various mixes of background load and CPU speed. The SAMiB algorithm has achieved a decrease of 14.93% in makespan time for 2000 jobs, proving the suitability of the multi-instance broker concept for the hierarchical cluster grid environment.