Auction based resource allocation in grids

Auctions have been used as the market mechanism for allocating resources to users in a grid. However, most of the existing work either consider only homogeneous resources, or pick a resource randomly for bidding to increase the number of jobs finishing within their deadlines. Random selection does not consider resource capabilities and hence, do not optimize other important metrics like the average turnaround time and the average budget spent per job etc. In this paper, we consider the resource allocation problem in a grid with heterogeneous resources. We present allocation policies using sealed-bid auction that reduce the average turnaround time and the average budget spent per job, while still maintaining a high number of jobs finishing within their deadlines. Simulation results are presented to evaluate the performance of the policies.