短事务、强实时双机容错系统的研究

在军事、工业控制以及电子商务系统中存在着大量的高可用,短事务、强实时应用,在这些应用中,采用双机系统具有较高的性能价格比,如何保证双机系统的强实时性,高可用度和服务“不断流”,是其中的关键技术难题,文中着重论述了系统可用度,故障检测,结果判别和状态切换中的关键问题,在理论的指导下,给出上实现策略和实际测试数据,测试数据表明本方案完全满足系统的要求,并且在具体工程实践中得到了应用,取得了明显的效果。

A Study of Short Transaction, Hard Real Time Dual Systems

There are many short-transaction applications, with requirements of high-availability and hard real time, in military, industry and electronic commerce. The dual systems are used in these applications with high performance-cost ratio. How to ensure the hard real time, the high availability and the "non stop" service are key problems in dual systems. A full solution, which can meet the requirements of the dual systems, is proposed in the paper. The system availability, fault detection, result comparison and state transition are emphasized in the paper. Firstly, the architecture of dual systems and the concepts such as physical processor, logic processor are defined. The system availability of dual systems with cut in success ratio is proposed also. Secondly, the fault classification is defined and the algorithms of "non fixed interval" fault detection, the results comparison, the system cut-in are proposed in the paper. These techniques can ensure the high efficiency, the synchronization between dual systems, and the report of faults rapidly. These means which can be applied to reduce the cut-in time are proposed with analysis. The solution has been applied in an important national project. At last, the test data fetched from that project are displayed in the paper, which also shows that the achievement of the solution in the project is outstanding.