Energy Efficient Scheduling of Servers with Multi-Sleep Modes for Cloud Data Center
Abstract
In a cloud data center, servers are always over-provided in an active state to meet the peak demand of requests, resulting in a large amount of energy being wasted. One of the options to reduce data center power consumption is to reduce the number of idle servers or switch idle servers to low-power sleep states. Energy Efficient Scheduling of Servers with Multi-Sleep Modes for Cloud Data Center However, when transiting to the active state, the servers can not process the requests immediately.
During the transition, there are delays and extra power consumption. We consider using state-of – the-art servers with multi-sleep modes in this paper. Sleep mode with smaller transition delays usually consumes more power when sleeping. We formulate this Energy Efficient Scheduling of Servers with Multi-Sleep Modes for Cloud Data Center problem as an integer linear programming (ILP) problem with millions of decision variables over the whole period of time. To solve this problem, we divide it into smaller-term sub-problems while ensuring feasibility and continuity of transition for each sub-problem through backtrack and update technique.
System Configuration
H/W System Configuration
Speed : 1.1 GHz
RAM : 256 MB(min)
Hard Disk : 20 GB
Floppy Drive : 1.44 MB
Key Board : Standard Windows Keyboard
Mouse : Two or Three Button Mouse
Monitor : SVGA
S/W System Configuration
Platform : cloud computing
Operating system : Windows Xp,7,
Server : WAMP/Apache
Working on : Browser Like Firefox, IE
Conclusion
We studied the problem of scheduling servers with multi-sleep modes for cloud data centers in this paper. The servers can make transitions for the sleep modes between one active state and different sleep states, which involves different sleep power and transition delays. We proposed a backtrack-and-up method for making the servers schedule, deciding how many servers in each state should be switched to which states in each tim eslot, so that the total power consumption can be minimized while meeting the QoS requirement.
