Energy savings when migrating workloads to the cloud
| Author(s) | Zheng, Yan | |
| Date Accessioned | 2017-12-01T15:22:16Z | |
| Date Available | 2017-12-01T15:22:16Z | |
| Publication Date | 2017 | |
| SWORD Update | 2017-09-05T16:29:46Z | |
| Abstract | It is well known that the data centers used by cloud service providers (CSPs) are among the most efficient data centers in terms of energy usage. Consequently, migrating workloads to the cloud can result in a decrease in energy usage. This paper presents results based on a large data set of over 40,000 machines (virtual and physical) spread across over 300 data centers. With this data we quantify the energy savings and the sources of the energy savings. We focus on lift-and-shift migration along with optimal cloud instance size selection, as this type of migration is relatively straightforward. The data indicates that this type of migration should reduce energy usage by an average factor of 4.5 and 7.8. Relatively little of the energy savings is from the efficiencies of CSPs data centers related to efficient cooling and lighting. Instead, most the savings are from using modern CPUs and by correctly sizing the instances so that systems are not underutilized. We also consider potential energy savings from refactoring applications to make use to auto-scaling. While such refactoring has the potential to achieve considerable energy savings, the savings are likely to be less than what is achieved by the initial migration to the cloud. These findings contract the popular belief that one needs to modify applications in order to achieve the benefits of the cloud. | en_US |
| Advisor | Bohacek, Stephan K. | |
| Degree | M.S. | |
| Department | University of Delaware, Department of Electrical and Computer Engineering | |
| Unique Identifier | 1013717835 | |
| URL | http://udspace.udel.edu/handle/19716/21749 | |
| Language | en | |
| Publisher | University of Delaware | en_US |
| URI | https://search.proquest.com/docview/1957944773?accountid=10457 | |
| Keywords | Applied sciences | en_US |
| Keywords | Cloud computing | en_US |
| Keywords | Data migration | en_US |
| Title | Energy savings when migrating workloads to the cloud | en_US |
| Type | Thesis | en_US |