Abstract
Nowadays corporate data centers leverage virtualization technology to cut operational costs. Virtualization allows to split the hardware resources of physical servers and to assign multiple virtual machines (VM) that run particular business applications to a single server. This has led to a new stream in the capacity planning literature dealing with the problem of assigning VMs with volatile demands to physical servers such that energy costs are minimized. The problem has been described as server consolidation and it seeks to leverage complementarities in the workload patterns to better utilize the servers. Live migration refers to a recent technology, which allows to migrate VMs to other servers at runtime, allowing for dynamic resource allocation, where a controller responds to server overload or underload and reallocates VMs aimed at maximizing energy effciency. This is often seen as the most effcient means to allocate hardware resources in a data center. Unfortunately, there is hardly any empirical or experimental evidence for this claim. In this paper, we provide the results of an extensive experimental analysis of both capacity management approaches on a data center infrastructure. We show that with typical workloads of business applications dynamic resource allocation does not necessarily increase energy effciency over well planned static allocation of VMs to servers and can even come at a cost, as migrations lead to significant overheads.
Original language | English |
---|---|
State | Published - 2013 |
Event | 23rd Workshop on Information Technology and Systems: Leveraging Big Data Analytics for Societal Benefits, WITS 2013 - Milan, Italy Duration: 14 Dec 2013 → 15 Dec 2013 |
Conference
Conference | 23rd Workshop on Information Technology and Systems: Leveraging Big Data Analytics for Societal Benefits, WITS 2013 |
---|---|
Country/Territory | Italy |
City | Milan |
Period | 14/12/13 → 15/12/13 |