Implementing the Data Center Energy Productivity Metric in a High-Performance Computing Data Center

摘要: As data centers proliferate in size and number, improving energy efficiency productivity has become an economic environmental imperative. Making these improvements requires metrics that are robust, interpretable, practical. We examine the properties of a number proposed productivity. In particular, we focus on Data Center Energy Productivity (DCeP) metric, which is ratio useful work produced by center to consumed performing work. investigated DCeP as principal outcome designed experiment using highly instrumented, high-performance computing (HPC) center. found was successful clearly distinguishing different operational states center, thereby validating its utility metric for identifying configurations hardware software would improve also discuss some challenges benefits associated with implementing efficacy making comparisons within among centers.