Paper Details

In cloud computing large number of data centers encompassing many servers are geographicallydistributed on the globe and connected in a network. These provide round the clock access to almost whole population onthe globe.To keep all the servers of a cloud operational a considerable amount of energy is consumed , which presentlyaccounts for 10% of the total operational cost and is expected to rise to 50% in the coming years.Statistically it has beenfound that on an average only 30% of the resources of a cloud are utilized at any given instant of time. Most of theresources therefore remain underutilized / unutilized. These resources which remain underutilized / unutilized consume alot of energy as an idle server consumes about two-third of the energy of the peak load. This is because of the fact thatthe servers must manage modules, disks, I/O resources and other peripherals in acceptable state. Therefore, a lot ofenergy gets wasted.In this thesis a green scheduling algorithm is proposed which penalizes low server utilization andfavors high server utilization. It concentrates the workload on a minimum set of servers and maximizes the number ofservers that can remain in idle state. Finally servers that are in the idle state are shut down and put in sleep mode usingDynamic Power Management. As a result, significant reduction in energy consumption is achieved

Idris Afzal Shah, Syed Arshid Ahmad Simnani, Hamid Hussain Haqani, Faisal Rasheed Lone, “Green Scheduling Algorithm for Cloud Centric Data Centers”, International Journal of Advance Engineering and Research Development (IJAERD), Vol. 5, Issue 13, pp. -, January 2018.