A Nonblocking Coordinated Checkpointing Algorithm for Mobile Computing Systems

A checkpoint algorithm for mobile computing systems needs to handle many new issues like: mobility, low bandwidth of wireless channels, lack of stable storage on mobile nodes, disconnections, limited battery power and high failure rate of mobile nodes. These issues make traditional checkpointing techniques unsuitable for such environments. Minimum-process coordinated checkpointing is an attractive approach to introduce fault tolerance in mobile distributed systems transparently. This approach is domino-free, requires at most two checkpoints of a process on stable storage, and forces only a minimum number of processes to checkpoint. But, it requires extra synchronization messages, blocking of the underlying computation or taking some useless checkpoints. In this paper, we propose a nonblocking coordinated checkpointing algorithm for mobile computing systems, which requires only a minimum number of processes to take permanent checkpoints. We reduce the message complexity as compared to the Cao-Singhal algorithm [4], while keeping the number of useless checkpoints unchanged. We also address the related issues like: failures during checkpointing, disconnections, concurrent initiations of the algorithm and maintaining exact dependencies among processes. Finally, the paper presents an optimization technique, which significantly reduces the number of useless checkpoints at the cost of minor increase in the message complexity. In coordinated checkpointing, if a single process fails to take its tentative checkpoint; all the checkpoint effort is aborted. We try to reduce this effort by taking soft checkpoints in the first phase at Mobile Hosts.