Statistical Service Assurances for Applications in Utility Grid Environments

In this paper we focus on techniques to support admissioncontrol and advance resource reservation for applicationsacquiring information technology (IT) resources fromutility grid environments. Utility grid environments offerprogrammatic access to resources for complex multitierapplications. In particular we consider business applicationswhich require resources continuously but that haveresource demands that change regularly based on factorssuch as time of day and day of week. We present a statisticaldemand profile approach for characterizing time varyingapplication resource requirements along with a methodthat provides statistical assurances regarding the numberof resources needed to satisfy the combined requirementsof many applications over time. We illustrate the feasibilityof our approach with a case study that uses resourceutilization information from 48 data center servers. Simulationexperiments explore the sensitivity of the assurancesto correlations between application resource demands andthe precision of the demand profiles.