SOA环境下分布式策略评估引擎研究

Study on distributed policy evaluation engine in SOA environment

为了满足面向服务体系架构中对大规模访问请求高效评估的需求,提出了一种分布式策略评估引擎XDEngine.该引擎的核心为策略分解模块和请求分配模块,具有分解策略和分发请求的功能.提出了按照主体属性进行策略分解的算法,该算法将策略分解成多个子策略,并使得部署到每个策略决策点的子策略的开销相等或近似相等.通过实验对比了XDEngine和Sun PDP中策略决策点的评估性能,测量了含有不同数量策略决策点的XDEngine的评估时间.实验结果表明:策略分解算法大幅度提高了XDEngine的评估性能,XDEngine的评估时间随着策略决策点数目的增多而降低;XDEngine能够满足面向服务体系架构中高效评估高并发量访问请求的需求.

A distributed policy evaluation engine termed XDEngine was presented to satisfy the requirement that a large number of access requests should be evaluated efficiently in the service-oriented architecture(SOA)environment.XDEngine contained two core modules,apolicy decomposition module and a request distribution module,with abilities of decomposing policies and distributing requests.A policy decomposition algorithm was proposed according to the subject attribute.The algorithm decomposed a policy into multiple sub-policies,and guaranteed that the cost of sub-policies deployed to each PDP was equal or approximately equal.Comparisons of the evaluation performance of PDPs in XDEngine with that of PDPs in the Sun PDP were made.Also,the evaluation time of XDEngine with different numbers of PDPs was measured.Experimental results show that the algorithm of policy decomposition improves the evaluation performance of XDEngine effectively.The evaluation time of XDEngine reduces with the growing numbers of PDPs.XDEngine satisfies the requirement that high concurrency of access requests are evaluated efficiently in the SOA environment.