Interplay of Virtual Machine Selection and Virtual Machine Placement
Previous work on optimizing resource provisioning in virtualized environments focused either on mapping virtual machines to physical machines (i.e., virtual machine placement) or mapping computational tasks to virtual machines (i.e., virtual machine selec
- PDF / 679,511 Bytes
- 15 Pages / 439.37 x 666.142 pts Page_size
- 63 Downloads / 259 Views
tract. Previous work on optimizing resource provisioning in virtualized environments focused either on mapping virtual machines to physical machines (i.e., virtual machine placement) or mapping computational tasks to virtual machines (i.e., virtual machine selection). In this paper, we investigate how these two optimization problems influence each other. Our study shows that exploiting knowledge about the physical machines and about the virtual machine placement algorithm in the course of virtual machine selection leads to better overall results than considering the two problems in isolation.
1
Introduction
As cloud data centers are serving an ever growing demand for computation, storage, and networking, their efficient operation has become a high priority. On one hand, the operation of data centers incurs huge costs and environmental impact. According to a recent study, data center electricity consumption in the USA alone will increase to 140 billion kWh per year by 2020, costing US businesses 13 billion USD annually in electricity bills and emitting nearly 100 million tons of CO2 per year [25]. On the other hand, servers often run with low utilization – in fact, a significant percentage of running servers do not do any useful work [1]. Virtualization has been widely adopted in data centers to consolidate workload on the necessary number of physical machines (PMs), with the aim of achieving high utilization and switching off unused PMs to save energy. For this purpose, virtual machines (VMs) are used as the virtual infrastructure for running the workload. Live migration technology makes it possible to migrate a running VM from one PM to another one without noticeable downtime. This way, data center operators can react to changes in the workload and always use the appropriate number of turned-on PMs to accommodate the active VMs, taking into account their current resource needs. However, too aggressive consolidation must be avoided because overloading physical resources leads to performance degradation. Furthermore, live migration of VMs incurs increased resource consumption, so that the number of migrations must be limited. c IFIP International Federation for Information Processing 2016 Published by Springer International Publishing Switzerland 2016. All Rights Reserved M. Aiello et al. (Eds.): ESOCC 2016, LNCS 9846, pp. 137–151, 2016. DOI: 10.1007/978-3-319-44482-6 9
138
´ Mann Z.A.
Optimization relating to the management of VMs has received considerable attention in the last couple of years because of its impact on costs, application performance and carbon emission [29]. As shown in our recent survey [21], most previous research efforts fall into one of two categories: VM placement and VM selection. The goal of VM placement is to determine a mapping of VMs to PMs with the objective of minimizing energy consumption while obeying performance constraints and keeping the number of VM migrations low [23]. On the other hand, VM selection is concerned with assigning computational tasks to VMs. VM selection
Task 0..*
Data Loading...
