Modeling Energy Savings for Job Migration in Grid Environments

An open issue to exploit Grid resources is related to job migration, possibly due to resource failures (real or perceived ones) or to the violation of QoS, it may be required to migrate an entire job, or parts of it, to other resources and reschedule the job there. There are several research questions associated with the migration concept. The two most crucial ones are determination of metrics and conditions when to migrate, and the selection of the resource to migrate the job to. In this work, we propose to combine job migration with energy consumption indicators. The objective is to understand if it is possible to exploit the migration of jobs also to save energy. Thus the main question of this work is "Can job migration save energy?". In order to provide an answer, we define a simple model to evaluate the energy consumption related to Grid job migration. An adequate Grid infrastructure is assumed to actually enable job migration, and the proposed model will take into account the overhead arising from that infrastructure.