The Green Grid’s Datacenter Metrics

By John E. West

May 1, 2008

At the beginning of April, HPCwire profiled The Green Grid, a relatively new organization focused on improving energy efficiency in datacenters and “business computing ecosystems.” After two years of operation, they have over 150 members that include power companies, hardware vendors, and end users. This group is poised to serve as the channel through which the IT industry starts to grow its management of energy use and environmental impact. On an energy per square foot basis, HPC’ers have some of the most significant challenges in an industry that the EPA estimates uses nearly two percent of the electricity consumed in the U.S. every year.

The old saw is that you cannot manage what you cannot measure, and The Green Grid has proposed two metrics for understanding datacenter efficiency: Datacenter Infrastructure Efficiency (DCiE) and Power Usage Effectiveness (PUE). These metrics are designed to give datacenter managers insight into how much power goes to actual computing tasks versus the power consumed in the datacenter as a whole (for cooling, lighting, and so on).

The metrics use the same quantities — one is simply the inverse of the other. PUE divides total facility power (TFP) by IT equipment power (ITEP); DCiE is the inverse of PUE. TFP is measured at the utility meter for the datacenter, while ITEP is measured from individual meters on circuits supplying servers, monitors, KVMs, network gear, storage, and so on. In order to get an accurate TFP, you should be sure to include equipment that supports the datacenter (and only the datacenter) but may not be located within the datacenter — for example, UPS equipment and chillers — as well as items that are often fed as “house power” — lights in the server rooms, for example — but which the datacenter usually doesn’t think about. Also, don’t forget to keep the IT equipment in the total figure for TFP. According to “Green Grid Metrics,” a whitepaper available at http://www.thegreengrid.org/gg_content/, ITEP should be measured “after all power conversion, switching, and conditioning is completed and before the IT equipment itself,” say at the output of the relevant PDUs.

PUE ranges from 1 to infinity (lower is better) and gives you a multiplier for considering an approximation of the real demands of equipment placed in your center. If you are considering adding a supercomputer that needs 1 MW, for example, and you have a PUE of 3.0, you’ll know that you’ll need to supply a total of 3 MW to run the system and all of its support components. This number is an approximation for the new system because the datacenter has fixed components that don’t vary with IT demand, such as lighting. But to the extent that these fixed components are small in proportion to the variable energy use items, PUE can be a good approximation.

The DCiE ranges from 1 to 100 percent (higher is better) and is a useful thumbnail for knowing what proportion of the total power consumed in your datacenter is used in the IT equipment itself. For example, a DCiE of 0.33 indicates that 33 percent of your power goes to IT gear.

So what is a “good” number and what does this mean in practice? The Green Grid doesn’t know yet. According to their metrics papers, there is research that indicates either PUEs of 1.6 or 2.0 should be achievable with careful design, but the recommendation right now is for datacenters to simply begin using and reporting the numbers so that the community can get a feel for the range.

To get a better handle for what this all means in practice I talked with Jim Smith, vice president of engineering for The Green Grid member Digital Realty Trust. From an IT perspective, the Digital Realty’s datacenters and mission look a lot more like supercomputing than enterprise computing. The $400M dollar company provides technology-related real estate (think turnkey datacenters) for customers around the world.

One of the insights that I got from Jim was the degree to which the metrics vary due to factors that aren’t under the datacenter manager’s control. My preconception simply reading the papers was that a datacenter would eventually automate these measurements and continually monitor them, responding to fluctuations and looking to them for immediate feedback on the effectiveness of energy efficiency initiatives.

In practice, it turns out that the datacenter metrics may be more useful as point measurements or as long term trends. Right now Digital Realty Trust evaluates the PUE primarily at datacenter commissioning, with load banks in the center to simulate a fully loaded datacenter.

“PUE and DCiE is indicative at commissioning because you’re at 100 percent load,” says Jim Smith. The commissioning load is a point event during which factors like server load and changes in climate and the outside environment are known. Digital Realty uses this commissioning assessment to measure the health of a new or recently reconfigured datacenter. Digital Realty is also beginning to explore the use of PUE as a contractual deliverable in new datacenter construction contracts.

In Smith’s experience, the metrics are highly sensitive to both server load and changes in the outside environment. Digital Realty’s centers in Dallas, Texas, and Dublin, Ireland, are good examples of the influence of climate. The climate in Dublin is very stable relative to Dallas, where temperature and humidity vary widely throughout the year. Smith indicates that in Dublin a day-to-day measurement of PUE might be useful, but in Dallas the metric can vary widely even within a single day with weather changes, making anything other than trend information over long periods of time or data from carefully controlled measurements not very useful.

The metric also varies with server load, which can fluctuate dramatically from hour to hour in a datacenter. “Higher loads are more efficient in terms of the metrics,” he says. This observation has lead the Digital Realty Trust to modularize their datacenters. For example, if you need additional capacity today and think that you’ll eventually need a 2.4MW datacenter, it is better to build three separate 600kW datacenters and grow into the new units as your needs grow than to build one 2.4MW datacenter today. This approach leads to more efficient cooling in the growth states leading up to full load. The modularized centers can be three discrete centers in different locations, or three units of what used to be one large continuous space physically and electrically divided.

Smith is measuring PUE at all of Digital Realty’s datacenters and is using those measurements to build a design database that he hopes to be able to use to gain insight into which combinations of computation support infrastructure technologies are most effective in datacenters of differing sizes in varying locations.

Despite the fact that the IT community is still in the early stages with PUE and DCiE, Smith emphasizes the importance of beginning to measure and track the quantities. “My one piece of advice is to get your measurement program going now because that’s going to be key to understanding what’s going on inside your datacenter, and it’s not that expensive,” he says. “Most people should have the ability to measure today.” A measurement program can start as simply as sending out a guy once a day with a clipboard to look at the meter. He also advises that datacenter managers should get hold of the power bill, and actually read it.

When you are ready to take things a step further, installing automated monitoring is not an expensive proposition. According to Smith, “In a $10M datacenter, a complete metering solution would cost between $40k and $60k.”

The biggest opportunity a manager is likely to have to improve the efficiency of operations is during a remodel, or during new construction when you can replace or install a highly efficient power and cooling distribution infrastructure. Almost everyone is going to have some significant remodeling in their datacenter over the next five to seven years, and if you start measuring these quantities now you’ll be in a much better position to make the changes that matter for your organization when that time comes.

Something else to think about? The specter of government regulation hangs over both U.S. and European datacenter operators. Jim Smith points to new EU regulations that go into effect next year requiring datacenters larger than 500kW to report their carbon footprint, which will also drive the need to jump start your monitoring program.

As you become more aware of what the metrics for your datacenter are, you are likely to want to start making changes to see what impact you can have. Smith explains that it is key at this point to tell everyone involved with the datacenter, from the users to the operations and maintenance people, what it is you are doing. “Taking this step gets a dialogue going and starts to get everyone involved,” and that’s when an efficiency agenda can really take off.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire