Improved Bandwidth Utilization based Check pointing Algorithm in Distributed Database

In this paper, we present an optimal-bandwidth, min-process coordinated check pointing algorithm suitable for network failure prone applications in distributed systems. In the developed algorithm, during normal computation message transmission, dependency information among clusters is recorded in the corresponding cluster head processes. When a check pointing procedure begins, the initiator from a cluster concurrently sends composite message to all the cluster head processes which after extracting individual messages from it, further multicasts individual messages to the corresponding currently active receiving processes in their corresponding clusters thus resulting in reduced transmission delay and communication cost, better bandwidth utilization and faster speed of execution. Quantitative analysis shows that proposed algorithm works efficiently in terms of better response time and maximum bandwidth utilization for applications running under critical conditions such as low bandwidth availability and thereby resulting in frequent disconnections.