Nowadays, there are known a number of systems, which are employed for the management of a set of various distributed computers and utilization of their idle CPU cycles. Each of these systems has made an important contribution to the development of distributed computing methods. At the same time, none of them solves all the problems that are essential for creation of full-fledged grid-infrastructures. In the paper the requirements for middleware, that is capable to integrate globally distributed separate non-clustered computers and complies with the grid standards, are defined. Also, the architecture of the job management system for such resource infrastructure is proposed.

grid, distributed computing, Peer-to-Peer systems, resource management

Programming, parallel computing, multimedia