Distributed Systems: Time-Reliability Trade-Off for Performance Analysis

Abstract

A distributed system consists of a collection of autonomous computers linked by a computer network and equipped with distributed system software. Distributed systems are implemented on hardware platforms that vary in size from a few workstations interconnected by a single local area network to thousands of computers connected via multiple wide area networks. This paper is an attempt to analyze the performance of the distributed system based on its reliability-time trade-off through an exhaustive search process. The Execution Reliability (ER), Communication Reliability (CR), Execution Time (ET) and Communication Time (CT) are considered in the study and represented in form of matrices namely ERM (,), CRM (,), ETM (,), CTM (,), respectively. These matrices are modified and manipulated according to the task combinations. The reliability for the distributed computing system is evaluated for each task combination along with the execution and communication Times. The optimal reliable solution for distributed computing system is obtained from the calculated reliability values for all task combinations. Finally, the Index value, a measure of performance optimization, is obtained.