Models
The modern high performance data center is a complex and costly operation. Integrating software, hardware, energy flow, and heat management technologies along with more efficient physical infrastructures, the Energy Smart Data Center (ESDC) project aims to drive supercomputer energy efficiency to unprecedented levels. To accomplish this will require new models with detailed understanding of the value proposition of every component. Integral to the Total Cost of Ownership (TCO) model will be energy and thermal management efficiency models such as Coefficient of Performance (COP), defined as IT equipment power divided by cooling power; Power Usage Effectiveness (PUE), defined as total facility power divided by IT equipment power; and Data Center Performance Efficiency (DCPE), defined as useful work divided by IT equipment power.
IT equipment power is clearly central to any data center model, but little data is available that characterizes the dynamic use of power by various components under varying loads and thermal environments. Facility designers typically use worse case scenarios that lead to over engineering. With detailed information under a range of loads, ESDC expects to be able to create models that will minimize the need to over design data centers.
Power is not the only aspect of a data center that is over engineered. Cooling capacity is based on Computational Fluid Dynamics (CFD) models of air and heat flow throughout the data center. The ESDC project intends to make accurate air temperature, velocity, humidity, and pressure measurements throughout the testbed facility to validate a CFD model under a range of conditions.
The goal of the project is to provide a real-time view of the testbed facility that will provide a complete view of energy and heat flow, supported by a model that will permit accurate assessments of alternative environments both physically and economically.