Many scheduling and load balancing solutions have been proposed for grid computing, but considerable parts of them do not clearly define the partners which are helpful in increasing system performance. We in this thesis study the partner selection problem which is an important problem in acquiring a good system performance in heterogeneous grid networks. The problem is formulated into four objectives: remain power (or node speed), queue length of jobs, link speed, and transfer time. The solution set of such formulated problem are fuzzy multi-objective decision making. The results are demonstrated by comparison between single objective and multi-objective decision making approaches in choosing a best partner.
Keyword: grid computing, heterogeneous, partner selection, fuzzy, multi-objective decision making
