摘要： 分布式计算技术提供了充分利用现有网络资源的有效途径。该文论述了基于解决生物计算中难解问题的具有开放接口的分布式并行计算系统的设计与实现技术。系统兼有开放式、异构性、容错性与易用性等特点。讨论了系统的容错性机制、检查点策略及任务调度算法。对Motif Finding问题的求解验证表明，分布式并行计算机制能大大缩短问题的求解时间，为计算领域的难解问题提供有效的解决途径。

关键词: 分布式计算系统, 容错性, 任务调度

Abstract: Distributed computing technology provides an effective way to adequately utilize network resources. This paper presents a universal distributed parallel computing system, which is based on solving NP-hard problem in computational biology and provides an open interface for client to upload application and call service. The system is characterized with openness, isomerism, fault tolerance and usability. The paper focuses on the fault-tolerant mechanism, checkpoint strategy and task schedule algorithm. Experiment for Motif finding problem shows that the system can shorten running time sharply. And it’s also an effective way to solve other hard problems in computing area.

Key words: distributed computing system, fault tolerance, task schedule

中图分类号: 

• TP309