Mapping Cooperating GRID Applications by Affinity for Resource Characteristics

The Computational Grid, distributed and heterogeneous collections of computers in the Internet, has been considered a promising platform for the deployment of various high-performance computing applications. One of the crucial issues in the Grid is how to discover, select and map possible Grid resources in the Internet for meeting given applications. The general problem of statically mapping tasks to nodes has been shown to be NP-complete. In this paper, we propose a mapping algorithm for cooperating Grid applications by the affinity for the resources, named as

MACA

. The proposed algorithm utilizes the general affinity of Grid applications for certain resource characteristics such as CPU speeds, network bandwidth, and input/output handling capability. To show the effectiveness of the proposed mapping algorithm, we compare the performance of the algorithm with some previous mapping algorithms by simulation. The simulation results show that the algorithm could effectively utilize the affinity of Grid applications and shows good performance.