Description

In large scale, i.e. geographically distributed data centers, they have huge number of servers. Because of the huge number of servers, the power cost is higher. The major problem in the geographically distributed datacenters is power cost and delay. In our system, we propose a optimization approach for scheduling a job and managing the server in the geographically distributed data centers. Decisions are taken to make the server active in random manner based on arriving jobs. Here we propose a two time scale approach for delay tolerant workloads. The decisions are taken to make the server active in a low time scale. Decisions to control the service rates of the server in faster time scale. Our proposed approach reduce the power cost and delay. In our proposed work we are going to use SAVE algorithm. The SAVE algorithm comprises of front end server, back end cluster and queue update. Based on this we predict the utilization of power in terms of execution time we evaluate the power cost. The request is being satisfied by very large scale, organically circulated data centers, each comprising a huge number of servers. Attitudes in the first category attempt to save power cost through power effective hardware design and engineering, which embraces conniving energy efficient chips, DC power supplies, and cooling systems. This work falls under the second class, where our goal is to deliver a unifying outline that allows one to exploit power cost saving openings across all these levels.