IT heavyweights plot The Green Grid

IT’s thirst for energy is growing exponentially, but the supply of reliable, clean power isn’t keeping pace. Thus companies are facing not only swelling utility bills struggling to keep their servers cool and humming, but, according to Gartner research, the threat that half the world’s data centers will run out of power by the end of 2008.

Thus eleven enterprise heavyweights, including AMD, APC, Dell, HP, IBM, Intel, Microsoft, Rackable Systems, SprayCool, Sun, and VMware — many of which are otherwise fierce competitors — have joined forces to form a non-profit group called The Green Grid, which will be officially announced Monday.

The group’s purpose: to take a holistic view of the computing ecosystem in order to address what it describes as an impending power crisis. The group’s charter calls for driving new user-centric metrics, technology standards, and best practices for curbing power consumption.

The self-described vendor-neutral consortium’s efforts already have born fruit. In conjunction with Monday’s announcement, The Green Grid will make available three white papers on its Web site. (The group gave me a sneak peek at the papers; details at the end of this entry.)

For now, the group is reaching out to other companies, as well as governmental agencies, in an effort to expand its roster and get interested parties involved with its various technical working groups. These groups will, among other things, be hammering out energy-efficient metrics, standards, and architectural specs for the long term.

While it might be simpler for just a handful of companies to develop standards, this collective approach is a must, according to Green Grid board member Bruce Shaw, director of worldwide commercial and enterprise marketing at AMD. “No one player owns the kingdom. It’s not just a processor problem. It’s not just a server problem. It’s not just a memory problem. Everyone in the ecosystem needs to be involved.”

By working together, he added, companies can develop ways for their respective greener technologies to “hook into one another,” thus increasing their overall effectiveness.

One of the group’s aims is to develop certification for data center gear that would be akin to the Energy Star seal found on kitchen appliances and the like, or the automotive industry’s CAFE (Corporate Average Fuel Economy) regulations.

Yet figuring out how to even measure power consumption for enterprise equipment in a meaningful way has been a significant challenge, notes Green Grid board member Colette LaForce, vice president of marketing at Rackable Systems. “Do you measure by performance per watt? Per dollar? Per thread? Something different?”

Though announced today, the consortium started work last April, with representatives for AMD, Sun, IBM, and HP getting the ball rolling. Not surprisingly, given the players involved and their numbers, there’ve been challenges from the get-go, including “getting these 11 companies to agree to a charter and figuring out how [intellectual property] would be handled.”

Membership to The Green Grid is open to end-users and technology companies alike. General membership costs $5,000 and includes access to all technical documentation and intellectual property licensing, voting privileges, as well as opportunities to attend events. Contributing membership, which costs $25,000, affords the aforementioned benefits and others, including eligibility to join working groups and review documentation.

As for the white papers that the group is unveiling on Monday, they’re respectively titled: “The Green Grid Opportunity,” “Guidelines for Energy-Efficient Datacenters,” and “Green Grid Metrics: Describing Datacenter Power Efficiency.” They do make for some interesting and informative reading.

“The Green Grid Opportunity”

This white paper spells out the case for developing energy-efficient standards and practices for the industry and outlines the group’s goals. It cites plenty of research on the topic, such as:

– “Traditional datacenters typically waste more than 60% of the energy they use to cool equipment,” according to Gartner research vice president Rakesh Kumar;

– “Commonly used power supplies have a typical efficiency of 65-70% and are a huge generator of waste heat, while units with efficiencies of 90% or better are available and pay for themselves over the life of the equipment,” according to Stanford University professor Jon Koomey;

– “Energy bills traditionally have accounted for less than 10 percent of an overall IT budget but soon could account for more than half,” according to Gartner research.

As for the group’s goals: “In the short term, the organization seeks to define a way for facilities organizations, IT managers, CIOs, regional power utilities, and governmental regulatory agencies to evaluate the performance-per-watt of the datacenter and their components … . While there are metrics today used to gauge the performance of the datacenter, their usefulness falls short when measuring datacenter performance-per-watt.”

“Long-term objectives … include defining a new datacenter power efficiency architecture to implement energy-efficiency policies natively through the instrumentation of devices.”

“Guidelines for Energy-Efficient Datacenters”

As its name suggests, this paper outlines some best practices for building a greener, more energy-efficient data center. It doesn’t go into deep detail on any one subject, but touches on topics such as system design (the physical placement of everything in the server room), which in and of itself can have a tremendous impact on energy consumption, according to the paper. “Two datacenters comprised of the same devices may have considerably different electrical bills. For this reason, system design is even more important than the selection of power and cooling devices in determining the efficiency of a datacenter.”

Additionally, it delves into several other topics, including cooling and even virtualization, which the paper notes “consolidates existing and expected future workloads. This reduces the number of physical servers required, thereby reducing floor space, cooling, and capital costs. It also increases the utilization of servers to improve energy efficiency.”

“Green Grid Metrics: Describing Datacenter Power Efficiency”

This is the certainly the most technical of the three white papers. In it, The Green Grid advocates, for the short term, two metrics for datacenter efficiency, Power Usage Effectiveness (PUE) — which looks at the ratio of “Total Facility Power” to “IT Equipment Power” — and Datacenter Efficiency (DCE), which reverses the ratio.

“The PUE and DCE provide a way to determine: opportunities to improve a datacenter’s operational efficiency; how a datacenter compares with competitive datacenters; if the datacenter operators are improving the designs and processes over time; and opportunities to repurpose energy for additional IT equipment.”

(Readers interested in greater detail should download the paper Monday when it goes live.)

The paper also proposes a long-term metric called the Datacenter Performance Efficiency (DCPE) — the ratio of “Useful Work” over “Total Facility Work.” The paper describes this metric as a natural evolution of the previous two: “In effect, this calculation defines the datacenter as a black box — power goes into the box, heat comes out, data goes into and out of the black box, and a net amount of useful work is done by the black box.”