A look at DOE's new Datacenter Profiling Tool

This past Friday I got a good look at the DOE's new DC-Pro (Datacenter Profiler) tool, while at the San Francisco Datacenter Dynamics conference. The overview and demo was driven by Bill Tschudi and Paul Matthew of Lawrence Berkeley National Labs (LBNL) who've been instrumental in helping DOE construct it as part of their Save Energy Now program (the beta release was back in June). What's really unique is that DOE, LBNL and the EPA have all collaborated to help create this, the first time I've ever seen the government take the lead ahead of industry in creating such useful IP for the private sector. An excellent presentation about this initiative is at the top of this DOE Page.

This tool absolutely complements the excellent work being done by the EPA to create an Energy Star rating for datacenters (where your facility will get the lable if it falls into the top quartile of facilities for that year), as well as datacenter metrics work being done by the Uptime InstituteThe Green Grid (of which Cassatt is a member).

The DC-Pro tool has (and will have) a number of useful outputs:

• Ability to help you track DCiE over time
• Outline of end-use energy & resource breakouts
• List of energy reduction potential (and idealized DCiE)
• Specifically define areas for improvement (i.e. power sources, HVAC, IT, etc.)

What's initially clear after logging-in to the web-based tool is the amount of depth/thought that went into the questions it asks, with context-sensitive pull-down menus, etc. It asks for so much data in fact, that I absolutely recommend that users download the checklist in advance to collect all the data they'll need to gather). Just to illustrate the though that goes into it, this tool even cares about your geography and zip code, because the carbon content of electricity varies for different parts of the country. You'd better be prepared to team up with your IT and Facilities counterparts to complete the data collection.

But what's also great is that the tool generates immediately-useful data, such as your DCiE/PUE, and how it ranks your measures relative to others who have used the tool so you can compare with peers (all of which has been anonymized so that no proprietary data is revealed). In this way you can also track how improvements affect efficiency for yourself and against peer groups.

Bill also outlined the roadmap for the tool (the tool's not entirely complete yet), which is impressive and agressive. Plus, by 2011, they'd like to see

• 3,000 data centers will have completed awareness training through classes or webcasts via DOE partners
• 1,500 mid-tier and enterprise-class data centers will have applied the Assessment Protocols and Tools to improve data center energy efficiency by 25% (on average);
  
• 200 enterprise-class data centers will have improved their energy efficiency by 50% (on average) via aggressive measures as accelerated virtualization, high-efficiency servers, high-efficiency power systems (e.g., fuel cells), optimized cooling, and combined heat and power systems
• 200 Qualified Specialists will be certified to assist data centers
  

Part of the roadmap for the tool are individual components that will make recommendations for improvements. Most of these sections are due to be integrated in the September '08 timeframe, and include the Air Management, HVAC, Power Chain, and IT sections. DOE and LBNL are reaching-out to external industry groups for input on these sections. It's my belief that these components will be highly comprehensive, and look towards some highly-agressive options & technologies that datacenter operators can leverage.

The montra I hear from everyone I speak with is "measure, measure, measure" (you can't control what you can't measure). But finally, someone has developed a tool in which to dump your measurements, and with which to compute your absolute and relative progress in becoming more energy-efficient.

Kudos!