Using mobile agents, it is possible to bring the code close to the resources, which is not foreseen by the traditional client/server paradigm. Compared to the client/server computing paradigm, the greater flexibility of the mobile agent paradigm comes at additional costs as well as the additional complexity of developing and managing mobile agent-based applications. Such complexity proposes the reliability issue; accordingly, reliability mechanisms for mobile agents must be established. In this paper, fault tolerance is used as a reliability mechanism. Various approaches exist for making mobile agents fault tolerance and their transactional support. These approaches have different strengths and weaknesses, and address different environments. Because of this variety, it is often difficult for the application programmer to choose the approach that best suited to an application. This paper introduces a classification of current approaches for fault-tolerant execution of mobile agents. The classification, which focuses on algorithmic aspects, aims at structuring of fault tolerant mobile agents and facilitates an understanding of the properties and weaknesses of the approaches. In addition, a non-blocking structure is presented for security of mobile agents. In this structure, the protocol that provides fault tolerance moves with the agent. This leads to implement fault tolerant mobile agents without changing their original plan. The costs of making mobile agents fault tolerant are compared against non-replication mobile agents. Furthermore, we show by increasing the number of stages and their size such overhead increases.
)
