Communication Pattern Based Checkpointing Coordination for Fault-Tolerant Distributed Computing Systems

This paper presents a new checkpointing coordination scheme which utilizes the communication pattern of the cooperating processes. In the proposed scheme, the checkpointing is coordinated for the limited number of processes based on the information regarding the communication pattern of the target program. Unlike the previous solutions which do not utilize the communication pattern, it is possible to reduce the coordination effort as well as the checkpointing frequency. Extensive simulation has been performed to evaluate the performance of the proposed scheme and we concluded that the proposed scheme significantly reduces the checkpointing overhead compared with the loose coordination schemes. 1. Introduction Checkpointing is an operation to save intermediate states of a process on the stable storage. With the periodic checkpointing, a process can roll back to one of the checkpointed states and resume the computation, when it fails. In a distributed environment, the computational stat..