Web服务失效处理的反射中间件技术应用与实现

Application of reflection middleware for services unavailability

为提高服务有效性,在Web服务的客户方和服务器方配置反射层,捕捉到导致服务失效的各种状态和参数,有针对性地从服务内部动态调整服务运行状态和配置。依据服务失效类型,给出客户方和服务器方反射层的处理方法和工作过程,并借助于分层混合专家网络(HME)作为服务失效的检测方案,基于极大似然的HME学习策略可以对高维、非线性和强耦合的状态空间进行学习和辨识。通过实验和数据分析,表明了基于HME网络的反射中间件可以高效地对服务失效进行检测和处理。

In order to enhance service validity, the reflective middleware deployed in client and server spot catches the trap which lead to unavailability of services, and modifies and configures the state of services from interior. The method and procedure are presented between the client and server based on the types of services unavailability. The expectation-maximization policy is introduced as a learning strategy in HME which as a detector can inspect and recognize state space that has high dimension, non-linearity and strong-couple features for service unavailability in the reflective middleware. An experiment and analysis show the reflective middleware based on HME can recognize and deal with services unavailability efficiently.