CAREER: Design and Implementation of Fault-Tolerant Distributed Computing Systems

该项目侧重于开发实现容错分布式系统的有效技术。尽管检查点和回滚恢复已经是在分布式系统中实现容错的已知技术，但设计有效的检查点和恢复协议所涉及的复杂性直到最近才得到彻底的解决和理解。在此理论基础上，(1)提出了开发有效的检查点和恢复技术；（ii）设立模拟测试平台，以评估新发展及现有的检查及回收技术的表现；(3)将我们的研究结果与现有的容错研究工作整合到研究生课程中。这项工作的重要预期结果是：(i)提高对构建可靠的分布式系统所涉及的问题的理解；(2)基于可靠的实验数据和坚实的理论基础，改进了分布式系统的容错技术；（iii）将我们的研究成果整合到分布式系统的高级课程中，这将有助于学生不仅理解构建可靠的分布式系统所涉及的复杂性，而且帮助他们获得构建此类系统所需的工具和技术。

This project focuses on developing efficient techniques for implementingfault-tolerant distributed systems. Even though checkpointing androllback recovery have been known techniques for achieving fault-tolerancein distributed systems, the intricacies involved in designing efficientcheckpointing and recovery protocols has been thoroughly addressed andunderstood only recently. Based on this theoretical foundation,(i) it is proposed to develop efficient checkpointing andrecovery techniques; (ii) implement a simulation testbed for evaluatingthe performance of the newly developed as well as existing checkpointingand recovery techniques; (iii) integrate the findingsof our research as well as the existing research work on fault-tolerance in thegraduate curriculum. The important expected outcomes from this work are:(i) an improved understanding of the issues involved in building reliabledistributed systems; (ii) improved techniques for fault-tolerance indistributed systems based on the hard experimental data as well as strongtheoretical foundation; (iii) integration of the results of our researchin an advanced course in distributed systems which would facilitatethe students not only understand the intricacies involved in buildingreliable distributed systems but also help them acquire the necessarytools and techniques for building such systems.