Resource Utilization-Aware Scheduling Technique Based on Dynamic Cache Refresh Scheme in Large-Scale Cloud Data centers

Abstract

In recent years, as researches on big data processing have become active, the frequency of use of HPC and large-scale data center is increasing for data processing. In addition, in the cloud computing environments, a large-scale environment consisting of huge nodes is being built, and additional problems arise due to the large-scale node configuration. In this paper, we analyze the overhead of gathering status information of large-scale nodes each time a virtual machine (VM) is migrated in a cloud environment composed of large-scale nodes. Then, we present a dynamic cache refresh (DCR) scheme to mitigate the unnecessary time of gathering status information of large-scale nodes. Our DCR scheme is able to obtain up-to-date information when necessary and reduce communication overhead without frequent cache refreshes. Based on experimental performance evaluations, we prove that our dynamic cache refresh (DCR) scheme achieves a significant reduction of the network bottleneck and the unnecessary delay time compared to existing cloud scheduler.