MobiGrid∗– Framework for Mobile Agents on Computer Grid Environments

This project focuses on the implementation of a framework for mobile agents support within a grid environment project, namely InteGrade. Our goal is to present a framework where time consuming sequential tasks can be executed using mainly the idle cycles of a network of personal workstations. The mobile agents may be used to encapsulate long processing applications (tasks). These agents can migrate whenever the local machine is requested by its user, since they are provided with automatic migration capabilities. Our framework also provides to the user a manager that keeps track of the agents submitted by him. 1. Motivation: Mobile Agents on Computational Grids The grid [Foster and Kesselman, 1999] paradigm was created to face two major problems of clusters and supercomputers: high prices and waste of resources. The computational grid idea was clearly inspired by clusters, in the sense that we have many computers interconnected by a network in order to provide together greater computational power. The idea is to provide computational resources similarly to the way we get power supply. The waste problem is addressed in a way that whenever a computer is idle, its computational power can be supplied to the grid. On this context, the idea of mobile agents can be interesting. They can be used to encapsulate opportunistic applications, which can use small slices of the available computational time of personal workstations, migrating to another machine whenever the local user requests his machine, always preserving the processing already done. On this way, mobile agents can be considered as a complementary tool to decrease even more the idle time of a grid. This text proposes our solution to provide a mobile agents environment in a grid and is organized in the following way: Section 1 explains the motivations to provide mobile agents support in a grid; Section 2 describes our project objectives; Section 3 gives a general overview of the framework; Section 4 provides ideas for future work and concludes this paper.