Performance Evaluation of Load Balancing in Hierarchical Architecture for Grid Computing Service Middleware
In this paper, we propose a hierarchical architecture for grid computing service that allows grid users with limited resources to do any kind of computation using grid shared hardware and/or software resources. The term limited resources includes disk or diskless workstations, Palmtops or any mobile devices. The proposed grid computing service takes into account both hardware and software requirements of the user computing task. Our grid system needs to maximize the overall system throughput, minimize the user response time, and allows a good grid resources utilization. On this aspect, we propose an adaptive task allocation and load balancing algorithm to achieve the desired goals. We have developed a simulation model using network simulator NS2 to evaluate the performance of our grid system. We have also conducted some experiments on our test-bed prototype. The performance evaluation measures confirm the good quality (grid saturation level close to 90% of the grid load) of our proposed architecture and load balancing algorithm.
Keywords: Distributed systems, Grid computing, Web-based applications, Load balancing, Performance evaluation
Download Full-Text
