Practical Fault-Tolerance for Mobile Agents

摘要: The amount of computational resources available on the Internet is increasing. Effectively using these for distributed computations challenging. An infrastructure called grids provides tools structuring and deploying large-scale Internet. One key problems in managing resources; based mobile agents are being advocated to solve this problem. However, be widely adopted, such must robust towards failures grid environment, thus require effective mechanisms agent fault-tolerance. To gain insight how applications perform Internet, dissertation investigates two master-worker algorithms, one group communication message flooding. Both algorithms executed simulations traces. results from running evaluating used infer requirements our fault-tolerance approach. This then derives a fault-tolerant protocol. protocol rooted primary-backup approach, where set backups monitor progress during computation. allows changed computation adapt current network topology. describes an implementation top platform, evaluates performance show that explicit management can beneficial performance, applicable outside scope computations.