ABSTRACT
Existing computational grid services are often organized in a remote or volunteer computing model. In the remote computing model, clients request services by sending data to servers; in the volunteer computing model, clients distribute their tasks to volunteer servers on demand. The former is limited to services that are predefined by the servers and the latter is unable to provide guarantees of service qualities from the perspective of applications. This chapter is devoted to a novel mobile agent-based push methodology to reverse the logic of pull-based task distribution in the volunteer computing model. It allows clients to dispatch their compute-intensive jobs as agents onto the servers so as to provide clients with customized computing services on the Internet.
