Recovery time improvement in the mobile database systems

Abstract

Mobile database system1 is a distributed system based on the client-server diagram. Recovery in mobile database systems is more complex compared to conventional database recovery because of an unlimited geographical mobility of mobile hosts. The mobility of these units makes it tricky to store application log and access it for recovery. The other hand, if compose both logging and checkpointing, cost of recovery is improveing and support independent recovery. Checkpointing is a time-consuming process because of in order to ensure global checkpoint consistency, all process should take checkpoint. In the way, some times lead to unnecessary checkpointing and exchange many messages. Obviously reduction in time of unification log or checkpointing, lead to reduce in recovery time. It is possible with collection list of dependent process's process that want to checkpointing and embed this list in mobile agent that moves with mobile host. This paper presents an log management and Low-Latency nonblocking Checkpointing Scheme, which uses a mobile-agent-based framework to reduce recovery time. Mobile agents very suitable for this scheme because of particular properties. In our scheme checkpoint algorithm forces a minimum number of processes to take checkpoints, which this processes is dependent processes. We compare the performance of our scheme with previous schemes and show that compared to these schemes, our scheme reduces overall recovery time. © 2009 IEEE.