GSM:一种多服务器系统的动态负载均衡调度模型

GSM:a Dynamic Load Balance Scheduling Model of Multi-Server Systems

设多服务器系统由 n个服务器结点和 1个中心任务调度结点组成 ,所有服务器具有同样的处理能力 ,任务的到达是一速率参数为 nλ的泊松流 ,任务的服务器处理时间服从参数为 1/μ的指数分布 (λ<μ) ,一次负载信息收集时间和任务调度时间在理想情况下可忽略不计 .基于这些假设 ,提出了多服务器系统的广义超级市场调度模型 (GeneralizedSuperm arket Model,GSM) ,详细分析并证明了该调度模型的解特性 .结论表明 :对于大规模多服务器系统 ,广义超级市场调度模型与随机选择服务调度策略相比 ,可指数级地提高平均任务消耗时间性能 ;与文献 [1,2 ,3]的结果相比 。

Suppose: ① a multi server system is composed of n service nodes and 1 central load balance scheduling node; ② all of the servers have the same processing capability; ③ the arrivals of tasks are a Poisson stream of rate nλ; ④ the service time for a task on a server is exponentially distributed with mean 1/μ, here λ<μ; ⑤ both load information gathering time and task scheduling time can be omitted under ideal conditions. Based on these assumptions, a Generalized Supermarket Model (GSM) is presented, and the solution properties of the model are analyzed and proved in details. The result shows that: by use of the Generalized Supermarket Model scheduling strategy for large scale of multi server systems, comparing with that of random service node choice strategy, the mean loss time for a task on a server can be exponentially decreased; and comparing with the results of literatures , it can reach the minimum of them.