Using network knowledge to improve workload performance in virtualized data centers
Abstract/Contents
- Abstract
- The scale and expense of modern virtualized data centers motivates running them as efficiently as possible. These data centers are primarily composed of physical machines running virtualization software, with each physical machine hosting many virtual machines (VMs) simultaneously. The performance of the workload running inside a VM is affected not only by other VMs on the same physical machine, but in the case of a workload that uses the network, the location of the VM and other VMs it is communicating with in the network topology, and the utilization of all network links in between. This thesis explores how the performance of workloads running inside VMs can be improved when network traffic and topology data informs the assignment of VMs to physical machines (VM placement). To answer this question, I built a network control system based on the OpenFlow control protocol, Beacon, and a cluster control system that managed the Xen virtualization layer, resource measurement, experiments, and the VM placement algorithms used to optimize experiments. This system is named Virtue, and was used to run and monitor experiments on an 80 server cluster. The VM placement algorithms evaluated in this thesis were able to improve the median performance of network-heavy, scale-out workloads by over 70% compared to random initial placements in a multi-tenant configuration. Performance improvements of 33% to 129% were observed when running the same experiment and varying both the edge link speeds between 100Mb/s and 750Mb/s, and the core network oversubscription ratio between 16:1 and 1:1.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2013 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Erickson, David |
|---|---|
| Associated with | Stanford University, Department of Computer Science. |
| Primary advisor | McKeown, Nick |
| Thesis advisor | McKeown, Nick |
| Thesis advisor | Kozyrakis, Christoforos, 1974- |
| Thesis advisor | Rosenblum, Mendel |
| Advisor | Kozyrakis, Christoforos, 1974- |
| Advisor | Rosenblum, Mendel |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | David Erickson. |
|---|---|
| Note | Submitted to the Department of Computer Science. |
| Thesis | Ph.D. Stanford University 2013 |
| Location | electronic resource |
Access conditions
- Copyright
- © 2013 by David Sterling Erickson
Also listed in
Loading usage metrics...