Is energy efficiency enough? Filling the engineering gap in data center design and operation

Abstract

The data center industry is focused on improving the energy efficiency of modern data centers due to their increasing energy costs and consumption, and the consequent carbon emissions. However, there is a lack of adopted process or metrics for data center cooling performance; this potentially puts facilities at risk from the unseen consequences of focusing on energy efficiency alone. The challenge for a data center operator is how to assess cooling performance. The facility's design likely makes assumptions about what IT equipment will be installed and how it will be configured that are not reflected in the operational configuration. This paper uses engineering simulation based on computational fluid dynamics (CFD) to show the relationship between the quantity of IT equipment that may be safely installed and the efficiency of the cooling system. Both are affected by changes to the cooling system settings and any IT equipment/applications deployed. The IT heat load and airflow requirement varies in legacy/enterprise style data centers and virtualized/cloud data centers alike, and this affects the cooling requirement. Using engineering simulation to predict the consequences of change provides a valuable complementary tool to the operator: it gives insight on how to avoid lost capacity, and helps them to better configure their facility when making decisions on infrastructure, IT deployment, or - in a virtualized environment - application deployment.