Integrating I/Os in Cloudsim for Performance and Energy Estimation

International audienceThis article presents an extension of the IaaS Cloud simulator CloudSim. This extension takes into account the processing of i/o workload generated by virtual machines within a data center, and evaluates the overall performance and energy consumption. Indeed, according to state-of-the-art mstudies, storage systems energy consumption may account for as much as 40% in a data center. So, we modified the time computation model of CloudSim to consider i/o operations. Additionally, we designed several models of storage system devices including Hard Disk Drives and Solid-State Drives. We also modeled cpu utilization to compute the energy consumptions related to i/o request processing. This was achieved through machine learning techniques. Our storage system extensions have been evaluated using video encoding traces. The simulation results show that a significant amount of energy, around 25%, is consumed due to i/o workload execution. This corroborates the soundness of our CloudSim extensions