Loading...

Saving Energy Consumption in Software-defined Networks

Dolati, Mahdi | 2015

1154 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 47866 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Movaghar Rahimabadi, Ali
  7. Abstract:
  8. consumption is an economical and environmental concern in nowadays world. Data centers (DC) are an important and energy power hunger part of modern life infrastructure. Many researches have shown that by considering special properties of DCs (e.g. redundancy) the energy consumption can be reduced. However new advances in hardware and software technologies have made DCs equipment more power proportional, i.e. they consume power proportional to their load. Unfortunately this new feature and many previous works do not have synergy. Also the performance of the DCs is so important that no DC owner is willing to trade it for lower power bills. Therefore algorithms that don’t consider the performance are of no use. The emergence of Software-Defined Networking enables efficient communication between network and its users. Users can gain extra information from the network to adjust their usage and help the network performance. In this thesis by considering the proportionality we will propose a joint solution to reduce DCs’ power consumption. Afterwards using SDN we have designed a mechanism to involve the users in the network control. These two solutions can be considered a complete response to the problem of energy consumption in DCs. Our optimization on average reduces the overall power consumption up to 15 percent and proposed virtual machine placement can reduce the network power up to 20%. Also our mechanism can utilize the network at desired threshold with 97% precision while keeping the delay constant and loss under 1%
  9. Keywords:
  10. Software Defined Networking (SDN) ; Data Center ; Energy Consumption Reduction ; Power Proportionality ; Network Awareness ; Network Robustness

 Digital Object List

 Bookmark

...see more