Checkpointing and Recovery Algorithms Using Mobile Agents on a Hamiltonian Topology

Traditional message passing based checkpointing and rollback recovery algorithms perform well for closely coupled systems. In wide area distributed systems these algorithms may incur large overhead due to message passing delay and network traffic. So to design checkpointing and rollback recovery algorithms for wide area distributed systems, mobile agents are introduced. Network topology is assumed to be an arbitrary hamiltonian graph. Processes are mobile agent enabled. When a process wants to take a checkpoint or is restored after fault, it just creates two mobile agents. The rest of the work of creating a consistent global state (CGS) for checkpointing or finding a CGS in case of recovery, is performed by the mobile agents. In the worst case, for concurrent initiations among processes, checkpointing algorithm takes mobile agent movements. A mobile agent carries size data irrespective of the number of processes.