Data centers are energy intensive, and opportunities exist to reduce energy use, but significant knowledge and skills are required to perform energy assessments. The Federal Energy Management Program’s (FEMP) Data Center Program assists federal agencies and other organizations with optimizing the design and operation of energy and water systems in data centers. It also funds the work at the Center of Expertise for Energy Efficiency in Data Centers (CoE) at Lawrence Berkeley National Laboratory. CoE develops and provides wide-ranging resources for managing data centers energy-efficiently. The main objective of this article is to provide an overview of the CoE’s Data Center Energy Efficiency Toolkit (Toolkit) and its free resources, including energy assessment simulation software, manuals, worksheets, templates, and listings. The Toolkit provides guidance on using the Tools in order to achieve energy efficiency and decarbonization in data centers. In our context, a “tool” refers to any resource for facilitating energy assessments. There is a FEMP on-demand webinar that provides additional information on the Toolkit: https://wbdg.org/continuing-education/femp-courses/fempodw140 A secondary objective is to increase awareness of the training courses offered by the Data Center Energy Practitioner (DCEP) program (http://datacenters.lbl.gov/dcep), which is partially funded by FEMP. These courses include training on most of the resources in the Toolkit.

Monitoring data center energy use at an aggregated level, as well as at the system level can provide data center owners and operators with critical insight that in turn, can help them reduce energy use. This report focuses on how monitoring data center servers can reduce energy consumption and enable more cost-effective operation. More specifically, this report reviews opportunities to leverage existing internal (onboard) server sensors for monitoring rather than external sensors for physical parameters such as intake air temperature and input power. Such an approach has many benefits for data center management, including higher data resolution and lower overall costs. Accessing onboard sensors is potentially a disruptive technology since hundreds of servers can be monitored without any external measurement equipment.

This report focuses on improving air management in small data centers due to the great potential for energy savings across the data center industry. To be able to implement air management, key environmental parameters such as IT equipment intake temperatures need to be monitored. This demonstration involves two inexpensive, portable measurement tools for assessing air management in small data centers on a limited, temporary basis. Access to simple, inexpensive tools for implementing and tracking air management is imperative in such environments. Besides evaluating the accuracy of the temperature measurements, this report also includes an in-depth evaluation of the ease of use of the tools.

This report addresses air-cooled computer server selection guidelines in order to facilitate energy efficiency and decarbonization in data centers. The objective is to help data center operators to operate more energy efficiently by purchasing computer servers that meet strict performance criteria.

Data centers are an important part of modern society, and most computer servers in these facilities are active and productive in storing, processing, and transmitting data. However, some are “zombies”. There are quite a few definitions of zombie servers, but a common thread is a server that consumes electricity but serves no useful purpose. That seems to encapsulate with a few words what a zombie does and does not do, although it still leaves room for interpretation. A zombie server, also known as a comatose or a ghost server, is often, but not always, a server that runs at very low utilization. Not only physical servers can be zombies, but also virtual servers. This document guides the important task of identifying zombie servers.

This demonstration, which took place at the U.S. Naval Research Laboratory (NRL) in Washington D.C., consisted of a commercially available high-performance computer immersed in the 3M Novec 649 liquid. The immersion cooling system was tested at a high computer load. Cooling for the bath was provided by a dry cooler located outside an HPC center at NRL. Two-phase immersion cooling using Novec 649 is not viable at this time. The primary obstacles that need to be overcome are IT equipment failures and costs. However, the demonstrated technology met the performance objectives for energy efficiency and greenhouse gas reduction. Before commercialization of this technology can occur, a root cause analysis of the failures should be completed, and the design changes proven.

A considerable amount of energy is consumed to cool electronic equipment in data centers. A method for substantially reducing the energy needed for this cooling was demonstrated. The method involves immersing electronic equipment in a non-conductive liquid that changes phase from a liquid to a gas. The liquid used was 3M Novec 649. Two-phase immersion cooling using this liquid is not viable at this time. The primary obstacles are IT equipment failures and costs. However, the demonstrated technology met the performance objectives for energy efficiency and greenhouse gas reduction. Before commercialization of this technology can occur, a root cause analysis of the failures should be completed, and the design changes proven.