Modeling and Simulation of Data Centers to Predict Behavior

Abstract

Data center modeling and simulation help to estimate and predict key parameters under certain a-priori known conditions. Data center systems must adopt end-to-end resource management, covering both cyber and physical components. This paper describes a systematic, easy-to-follow approach for data center modeling and simulation using a cyber-physical systems lens. The methodology is aimed to facilitate the estimation of quality of service, airflow and power requirements, and key indicators to assess data centers, and to then compare them to one another, or compare different scenarios under which data centers operate. The results contribute to communicate and better understand data center behavior and areas of improvement.