A Request Distribution Strategy Based on Static Time Interval

Cluster has become a new generation of distributed system. Among the server cluster technologies, request distribution technology is used to distribute requests among server nodes, providing support to large-scale concurrent access to server cluster. In order to overcome the weakness of existing request distribution strategies for server cluster in average response time and computation cost, this paper proposes a request distribution strategy based on static time interval. Its basic idea is that it divides the update interval into several subintervals and introduces randomness into the selection of server node for the requests arrived in a subinterval. We give an implementation of the strategy and use MATLAB as the simulation platform to carry out simulation experiments in testing the performance of the strategy. Our theoretical analysis and experimental results show that, the strategy we propose achieves shorter average waiting time, more convenient implementation and lower computation cost than other several existing request distribution strategies.