A genetic algorithm perspective to distributed systems design

Distributed computing systems (DCSs) have the potential for high reliability. When the topology of a DCS is fixed, the DCS's reliability depends mainly on the allocation of various resources. One of the important resources to be allocated in a DCS are various files. In this paper, a genetic algorithm (GA) based file allocation and design scheme has been developed for distributed systems. In this scheme, files are allocated to the nodes of the DCS such that the overall effectiveness of executing a program, which requires files from remote nodes, is maximized. Different topologies are considered to illustrate the GA-based solution approach. The effectiveness of the GA-based approach is demonstrated by comparing the results obtained using the GA with the results obtained by an exhaustive search of the problem state space. A timing study that was carried out also illustrates the effectiveness of the GA-based approach. This paper also studies the impact of the probability of crossover (Pc) and the probability of mutation (Pm) on the results obtained using the GA.