Matchmaking on Clouds and Grids

Abstract

Clouds and grids are composed of distributed heterogeneous resources that must meet various QoS and performance requirements. The performance of these systems often depends on the matchmaking strategies used by the resource management middleware to allocate requests to resources. Matchmaking on these systems can be difficult when newly available resources are brought into the system since the matchmaking middleware will have no knowledge of the scheduling policy used at these resources. This paper presents a matchmaking strategy that does not use detailed knowledge of the scheduling policies used at resources. This allows the system to incorporate newly available resources in order to increase system performance. Results of performance simulations for the strategy are very encouraging.