Placement of replicas in distributed systems using particle swarm optimization algorithm and its fuzzy generalization

摘要

Quick data access is an essential issue in large distributed systems. When distributed systems grow in size user access time increases, thus replicating data objects onto different servers can play an important role in distributed systems performance. Replica placement in distributed systems can be considered in dynamic and static environments. In this paper, we study the replica placement problem in the case that write/read demands are static. Particle swarm optimization algorithm and its fuzzy generalization are evolutionary paradigms to face optimization problems based on swarm intelligence. In this paper, PSO capability and its fuzzy generalization to effectively face replica placement problem in distributed systems is examined. We evaluate our proposed methodology results, comparing performance with two evolutionary algorithms along with a greedy paradigm which are previously reported as successful techniques to solve discussing problem.