Data Placement of Scientific Workflows based on Nature Inspired Algorithms on Cloud

Nowadays, there are big data which are transferring among data centers on cloud environments. These data consume network bandwidth, storage resources and may use up some computations. These resources are distributed geographically on the cloud environment. Since the policy of data centers differs from each other in the cloud environments, the utilization of them would certainly cause significant impact on the costs that the end users have to pay for it. Each task needs some input datasets and these data must be available at the computation host. There might also be some tasks which need the outputs of other tasks on the other data centers, as their input datasets, so the data need to be retrieved from the other resources. Therefore, the data movements would be inevitable among data centers in the cloud environment. The nature inspired algorithms provided a solution for solving complex problems resulting in outstanding outcome. In this work, the firefly algorithm and an extended genetic algorithm have been utilized to produce the optimal data placement and their outputs are compared to random data placement. The result shows about 15% improvement against random data placement.