Price/cooling aware and delay sensitive scheduling in geographically distributed data centers
Geographically distributed data centers of cloud provider incur heavy electricity costs due to high prices as well as inefficient workload management among the data centers. To bring down the operational costs dynamic power management is used as the basic approach where different power modes exist for a server. However the air flow pattern of the cooling systems is not taken into consideration in the existing works. This paper investigated that the power consumption and hence the electricity costs of the active servers in the data center is influenced by server utilization as well as output temperature of the cooling unit and proposed two algorithms, Electricity Cost Saving Workload Management Algorithm(ECSWMA) and Electricity Price Aware Workload Management Algorithm (EPAWMA) that jointly manage the workload of all the data centers run by a cloud provider in a cost effective manner. Experiments show that the proposed algorithm lowers the Accumulated Electricity Costs of the active servers to a large extent.