基于随机流模型的校园网格存储服务分析

Stochastic Fluid Models Analysis for Campus Grid Storage Service

校园网格存储服务是汇聚校园高性能计算中心、各个院系或研究所的服务器以及单个PC机贡献的存储资源,建立三级存储架构,为用户提供数据存储的注册、分配、调度、释放服务。由于节点的动态性,导致所提供的存储空间总数动态变化,为研究存储服务性能,采用随机流模型进行分析。它是一种宏观上对高速网络进行有效分析的模型,是基于排队论模型分析的一种替代方法。通过该模型分析校园网格存储服务性能,得出节点请求分配概率随节点数目变化的关系。

Campus grid storage service is to aggregate the storage resources in the servers of high performance center and colleges or departments （Institutes） in university, and the storage resources in Personal Computers （PC） to provide the services of register, allocation, scheduling, free allocation for uses through building three level storage architecture. Due to the nodes＇ dynamites, the total storage space that nodes contribute will dynamically change with time. To study the performance of the storage service, the stochastic fluid model is adopted. The model is an effective analytical model macroscopically for highspeed networks, an alternative model to queuing-based model. We use the stochastic fluid model to analyze the campus grid storage service, and the function between the storage allocation probability and number of nodes is got.