Energy-Efficient and SLA-Aware Virtual Machine Selection Algorithm for Dynamic Resource Allocation in Cloud Data Centers

摘要

Energy consumption constitutes a significant proportion of data centers' operational costs. Furthermore, the establishment of large scale Cloud data centers due to the fast growth of utility-based IT services made the energy usage of data centers a concern. Cloud data centers use load balancing algorithms to allocate their physical resources (CPU, memory, hard disk, network bandwidth) efficiently on demand and hence optimize their energy consumption. In the load balancing process, some Virtual Machines (VMs) are selected from over-or under-utilized physical hosts and these VMs are migrated, while live and running, to other hosts. This live migration can result in Service Level Agreement Violations (SLAVs) and consequently low Quality of Service (QoS). Thus, in this paper, we propose an energy aware VM selection policy to minimize the number of migrations and consequently decrease SLAVs. Load balancing has three stages: a) Detecting over-and under-utilized hosts; b) Selecting one or more VMs for migration from those hosts; c) Finding destination hosts for the selected VMs. The focus of this research is on the VM selection stage of CPU load balancing. Our proposed VM selection algorithm considers CPU utilization of the VMs on each host and any linear correlation between the CPU usage of the VMs. The algorithm was evaluated on two different real Cloud data sets provided by the CoMon project and Google. Its performance was compared to our benchmark policy that only considers minimum migration time for VM selection. The results showed that our proposed algorithm decreases SLAVs by 66%, ESV (SLAVs × energy consumption) by 64% and the number of "re over-utilized" hosts by 81% when the CPU usage of VMs in a data set are highly correlated (e.g., as in the Google data set).