基于"多层日志"的工作流的失败处理

工作流管理技术是实现企业信息集成的重要使能技术。它控制、协调分布于企业范围内的多个计算节点上具有异构、自主性等特点的应用系统，以共同完成企业的业务目标。工作流系统支撑着企业关键业务的运行，因而，保证工作流管理系统在失败情况下可靠运行显得日益重要。本文从企业应用系统具有的异构、自主、分布特点出发，详细分析了工作流基本失败的原因，并提出了层次型工作流失败模型。在此基础上，通过扩展WfMC体系结构，采用了一种新的建立“多层日志”的失败处理策略，使工作流管理系统在失败情况下，能恢复到失败前的“最近一致状态”，并从此状态前向执行、推进工作流实例，从而减少了有用工作的丢失，也提高了工作流系统的可靠性。

Multi-level Log Policy-based Handling Failure in Workflow Management System

The workflow management technology is considered as an enabling technology for implementing Computer Integrated Management System (CIMS). It coordinates and controls the heterogeneous and autonomous application systems that are physically distributed in various network nodes in order to achieve business goal. Workflow management system provides the support for enacting key business processes of an enterprise. Therefore, it's especially important to ensure the correctness and reliability of workflow execution in the presence of failure. In this paper, the reason for the workflow failure is analyzed and a hierarchy failure model is presented. Based on the model, a multi-level log policy for handling the failure is adopted, which will bring workflow system to the mostly-recent-consistent-state before failure occurs and continue to navigate the business process from the consistent state. As a result, more useful works are saved and the performance of workflow management will be improved.