We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Virtual Machines Online Acquisition.
- Authors
Alouane, N.; Abouchabaka, J.; Rafalia, N.
- Abstract
Clouds basically offer a set of instance acquisition solutions, it's either an on-demand plan where the user has to pay the full VM hourly pricing or can go with a commitment for a X duration, then the user can benefit from a Y percent of reduction over the total VM reservation period. That point of shifting or decision making becomes more difficult during the last couple years, with this big number of service reservation offers with various durations that we have on the market today and knowing the fact that not all workloads are easy to predict, it forces the user to think about an optimal combination of these offers, while maintaining the same availability level, consistency and latency of the on-demand solution. In this paper, we introduce two deterministic algorithms for the multi-slope case, that incur no more than 1+1/1-α and 1+2/1-α respectively, compared to the cost obtained from an optimal offline algorithm, where α is the maximum saving ratio of a reserved instance offer over on-demand plan. Our simulation driven by the google cluster usage data-trace shows that more than 30% of cost savings can be achieved when applied to a real cloud provider like amazon web services, while 40% when purchasing instances through a cloud broker service.
- Subjects
VIRTUAL machine systems; AUTOMATED library acquisitions systems; COMPUTER users; LOGICAL prediction; SIMULATION methods &; models
- Publication
IAENG International Journal of Computer Science, 2018, Vol 45, Issue 2, p304
- ISSN
1819-656X
- Publication type
Article