Tasks Distribution Strategy based on Cluster in MWfSCC

Cloud computing technology offers a new way to develop workflow system. Migrating workflow system based on cloud computing paradigm (MWfSCC) can obtain high system performance and service quality in cloud computing environment, where an agent (called CMI) is deployed to fulfill several tasks by migrating to other computers (called work machines). In general, task execution involves several input data sets usually placed on distributed work machines. On the other hand, task execution will also generate some data sets, which may be regard as input data sets for other tasks. Traditionally, CMI moves to work machines one by one for the reason that tasks execution places are also distributed. In this way, it is very important to arrange the tasks on different work machines in order to reduce CMI’s movements. In this paper, we propose a tasks distribution strategy based on cluster in MWfSCC, which contains tasks classifications and arrangements. Simulations show that tasks distribution strategy based on cluster can effectively reduce CMIs movements during workflow execution, and hence can significantly improve the performance of MWfSCC.