基于OGSA的网格服务容错框架的研究和应用

开放网格服务体系结构(OGSA)中的服务容错需要可扩展的、灵活的容错框架,能够支持容错策略与应用程序的分离.针对这一需求,提出了一种基于OGSA的层次式容错处理框架.框架能根据不同服务质量需要灵活设置故障处理策略,选择不同容错机制;在此框架下,实现了基于服务实例池的热备份以及基于服务检查点的服务重构等容错机制.该容错框架及机制应用于HUSTgrid平台,较好地满足了网格服务的容错需要,并在应用系统CoGIS中得到了验证.

Research and application of fault tolerance framework for grid service based on OGSA

Open Grid Services Architecture(OGSA) requires scalable and flexible fault tolerant mechanisms to address different requirements such as support for diverse failure handling strategies and separating failure handling strategies from application codes.We propose a hierarchical framework for fault tolerance based on the OGSA.In this framework,diverse strategies and mechanisms can be flexibly configured to meet different requirements of fault tolerance of services.Implementations of several mechanisms are presented and described in detail,including service instance hot standby in instance pool and service recovery based on checkpointing etc.This framework and the mechanisms are successfully implemented in HUSTgrid.Experiment results in CoGIS,which is an application deployed in HUSTgrid platform with multi-fault-tolerance mechanisms requirements shows that our fault tolerance framework and the related mechanisms work efficiently.