Novel dynamic shadow approach for fault tolerance in mobile agent systems

Mobile agents are the processes that can migrate from one machine of a system to another machine in order to satisfy requests made by their clients. Since mobile agent moves from one server to another server in an itinerary it is easily prone to various faults like server crash or agent crash, which makes fault tolerance one of the main issues for reliability in information retrieval applications. In this paper we have also proposed a new fault tolerance approach to tackle the problem of server crash by creating a cloned copy of the original agent. The clone follows the original agent in the itinerary so that in case of failure of original agent, it can be recovered from its clone. We have also added checkpointing, so as to limit the roll back of the agent in case of failure of both original agent and its clone. We have implemented the proposed approach on the Aglets mobile agent system and evaluated in terms of parameters such as total trip time, successful migration time, checkpoint time. The results show the improvement in the performance by improving total trip time.