Development of a Liquid Cooled Rack Specification

About the Project

Project Background & Purpose

There are currently no common specifications with most liquid cooling solutions being unique and proprietary. Such lack of standards and lack of multi-source solutions are seen as impediments to wide market adoption of liquid cooling. Influential market players could drive faster development and adoption of such technology for the benefit of themselves as well as the industry. Many vendors have entered the market with liquid cooling solutions, however most/all are proprietary and generally non-compatible with each other. This is acceptable for some homogeneous compute environments, but less acceptable for non-homogeneous compute environments where standard multivendor solutions are strongly preferred. Buyers in the market would benefit from greater standardization. The hypothesis is that incompatible proprietary systems are a market barrier that will continue to inhibit adoption of warm liquid cooling.

The scope of this project is to develop an open specification for the secondary fluid (closed loop between the CDU and the IT equipment), manifolds, tubing, quick connectors, and the operating conditions. It does not include the CDU or the heat exchangers in or on the IT equipment. The goal of the project is to develop a liquid cooled rack specification that could accommodate multiple vendors and provide a reusable infrastructure for multiple refresh cycles with a variety of liquid cooled servers/suppliers.  The specification could include but is not limited to, fluid selection and quality, supply pressure, temperature, and flow, delta pressure and temperature, header size and material, connection spacing, size, and details. Ideally the specification would be compatible with server rack designs promulgated by multiple open standards organizations such as Project Scorpio (China), the Open Compute Project (OCP), and Open19. 

Working Group Members

The U.S. Department of Energy (DOE) and China’s Ministry of Industry and Information Technology (MIIT) are supporting this bilateral initiative to increase efficiency in data centers. The Lawrence Berkeley National Laboratory (LBNL) and China Institute of Electronics (CIE) steer this initiative and coordinate the contributions of industry stakeholders. A small working group of influential potential users/buyers of warm liquid cooled equipment that are active in open standards organizations was formed. The working group includes staff from the following organizations:

  • Lawrence Berkeley National Laboratory
  • China Institute of Electronics
  • Tencent
  • Baidu
  • Alibaba
  • Intel
  • Facebook
  • LinkedIn
  • Google
  • Microsoft

Project Progress & Focus

The Open Specification for a Liquid Cooled Server Rack Progress Update document details the accomplishments made by the working group so far. The working group has focused on the following specifications:

  • The wetted material list (all components must be compatible with this list)
  • Water based transfer fluid quality and treatment
  • Proposed fluid operating ranges
  • A universal (multi-vendor) quick connect
  • General operating specifications (such as pressure and temperature ranges)

Challenges

The proprietary nature of both the chemical compositions of water based transfer fluids and the quick connects.

Relevant Documents

  • Open Specification for a Liquid Cooled Server Rack Progress UpdateThis document is a progress update on the development of an open specification for a liquid cooled server rack, including the proposed wetted materials list, wetted materials to be avoided, proposed fluid operating ranges, fluid thermo-physical properties, fluid quality, and other considerations. 
  • Presentation- Harmonization of Open StandardsSlides from presentation given at the Data Center Dynamics Conference in San Francisco in June, 2018. The presentation covers the working group effort to develop a liquid cooled rack specification. 
  • CoE Liquid Cooling PageLiquid cooling is valuable in reducing energy consumption because the heat capacity of liquids is orders of magnitude larger than that of air and once heat has been transferred to a liquid, it can be removed from the data center efficiently.