Finally, proof positive that PUE is garbage.

Vern Burke, SwiftWater Telecom
Biddeford, ME

I’ve just been reading a piece about Microsoft removing fans from their data center servers and that having a negative effect on their PUE numbers. I’ve written on this blog before about the problems with PUE, now we have proof that it needs to be put out of it’s misery.

In a nutshell, PUE is the ratio of power consumed by the IT equipment of the data center, versus the entire power consumed by the data center. A PUE of 1.0 would indicate a data center where all the power is being consumed by the IT equipment. A PUE greater that 1.0 indicates a data center where a certain amount of power is being consumed by other than IT equipment, the biggest chunk of which is cooling.

The problem I’ve written about before with PUE is the failure to take into account the actual work being accomplished by the IT equipment in the data center. Throw in a pile of extra servers just simply turned on and idling, not doing anything useful, and you’ve just gamed your PUE into looking better.

The problem shown here is even more damning. Microsoft determined that data center energy consumption could be reduced by removing the individual cooling fans from its servers and increasing the size of the data center cooling system. Since the increase in power for the data center cooling systems is less than the power required for the individual server fans, the data center accomplishes the same amount of work for less total energy consumption, an efficiency win in anyone’s book.

The side effect of this is that, even though the total energy consumption for the data center is reduced, transferring the energy usage from the fans (part of the IT equipment number) to the cooling (part of the non-IT equipment number) makes the PUE for the data center look WORSE.

Gaming the metric simply made it inaccurate, which was bad enough. Any efficiency metric that shows a net gain in data center efficiency (same amount of work accomplished for less energy consumed) as a NEGATIVE is hopelessly broken. This also has the side effect of making a mockery of the EPA’s Energy Star for data centers, since that award is based directly on the data center’s PUE.

Misleading, inaccurate, and now totally wrong, this sucker needs to go where all the other bad ideas go to die.

SwiftWater Telecom Green Eco Cabinet Filler Panels, insulated, lightweight, inexpensive

Thursday data center tidbits: boneheaded data center efficiency statistics and more!

First up today, we have a piece about the EPA’s Energy Star data center rating system. Proving the old saw about lies, damn lies, and statistics, we get:

“Of the numerous variables analyzed, only the annual IT energy consumption was found to be statistically significant in explaining the variation in energy use; data showed that facilities with higher IT energy consumption have lower PUE values, on average. The final regression model predicts a PUE value based on the IT energy.”

This has to be the most ridiculous thing I’ve ever heard. Data centers with high power consumption are expected to get better PUE? Based on what?

So, not only is the measurement a crock (PUE), the target the data center is measured against is conjured out of thin air and about as substantial.

Next is from a piece asking “Is virtualization cloud computing?“. We get this jewel from this reading:

“To continue this analogue, cloud computing can happen without virtualization. Certain hardware, operating system and even application clusters can deliver cloud services. ”

Cloud computing is an extension of virtualization, it’s not possible to separate the two and still have something meaningfully labeled as cloud computing. Sorry, go sit in the corner with everyone else who misuses the term cloud.

Email or call me or visit the SwiftWater Telecom web site for cloud computing services.

Vern

The green data center, the EPA, Energy Star, PUE, and total cluelessness.

I’ve just been reading a post about the EPA’s new Energy Star rating for data centers. It’s astonishing when you see the EPA having this big a bozo attack.

The first item on the list is the use of PUE (power usage effectiveness) as the criteria for awarding the Energy Star rating (I have no idea what happened to the EPA’s EUE metric they were trumpeting just a few months ago). In simple terms, PUE is the total amount of power used by the data center vs the amount of power used by the IT equipment in the data center. The problem with this approach is that it doesn’t take into account the actual amount of work being accomplished for that power, so it actually encourages the deployment of idle and underutilized equipment to skew the PUE metric for publicity and marketing purposes. Not very green at all.

PUE can be useful for evaluating the effect of changes made to a single facility, but it’s worse than useless for trying to compare multiple non-homogenous data centers. It’s so easy to skew that any comparisons are not only misleading, they’re downright dishonest as well.

The really astonishing thing is to hear the EPA Energy Star manager claim that the Energy Star rating didn’t have to take into account climate differences due to data center location because they didn’t have any statistical effect on PUE. This is the most clueless thing I’ve ever heard.

I’m sure all of these major companies that have located data centers in cooler climates to take advantage of free air cooling to drastically reduce electrical consumption required to cool the data center will be shocked to hear this (free air cooling circulates cool outside air to remove heat without any requirement for chillers). All of the massive power savings trumpeted in the news lately must just be a myth. Keep on mind, reducing cooling power usage reduces the overall power usage of the facility, driving down PUE.

Even for legacy cooling systems, cooler outside air increases efficiency. The larger the difference in temperature, the easier it is for the cooling system to dissipate or sink the waste heat. This means less energy involved in removing the same amount of heat.

To say that these techniques have no statistical bearing on PUE flies in the face of the laws of physics and just plain reality. This isn’t just some theory, it’s being done and it’s being proven out.

I wrote a week ago that this process was going to be a real circus and it’s nice to see them bearing me out. Alexandra Sullivan, EPA Energy Star manager, and the EPA as a whole get an en mass award for Data Center Bozos of the Week. The bulk shipment of multicolored wigs and red squeaky noses is on the way.

Call or email me or visit the SwiftWater Telecom web site for hype (and bozo) free green data center and cloud computing services.

Vern

Monday data center tidbits.

First up today is a piece about companies not giving their employees incentives not to do inefficient things in the data center. Incentives for not screwing up? That used to be called keeping your job.

Next is a piece about data center Energy Star ratings based on PUE coming in June. I’ve written before about how PUE is useless as a tool to compare different data centers and is ridiculously easy to manipulate. Let the shoveling of BS commence!

ps I notice the Energy Star process only talks about self entered data and no audit process to verify the data center is actually doing what they say they are. This should be entertaining.

Are you looking for cloud computing services, virtual servers, workstations, desktops, or file servers? Call or email me or visit the SwiftWater Telecom web site today!

Vern

Wednesday data center tidbits.

First up this morning, we have an article quoting a Microsoft person pondering about why many data center operators havn’t adopted any of the current data center “efficiency” metrics. It’s simple, the Green Grid’s PUE (power usage efficiency) and the EPA’s EUE (energy usage efficiency) are both a crock, since they’re ridiculously easy to game. Any data center energy efficiency metric that measures anything except total energy consumed vs total useful work achieved is misleading at the very best.

Vern, SwiftWater Telecom

aurora resilient cloud service

Monday data center tidbits.

First up on the list today is a story about the University of Illinois building a new supercomputer data center with no power backup. Apparently they consider their power grid to be super reliable so they don’t need power backup. Good luck with that one.

Next up is the story that the EPA has announced the finding that green house gases endanger the public health. If you’re not actively moving to get your data center greened, you better get your keister in gear. Oh, and if you’re still running a massive water sucking cooling system, better look for a replacement, the EPA points to serious upcoming water availability problems too.

Vern, SwiftWater Telecom

Aurora RCS virtual cloud computing

Finally, smart efficiency measurement for the data center.

After having PUE and EUE foisted on us, I knew someone was eventually going to get it right. This article I just read is spot on for the right way to measure data center efficiency.

First, I’ll take a brief look back at The Green Grid’s PUE (power usage effectiveness) and the EPA’s EUE (energy usage effectiveness). PUE measures the ratio of power coming into the data center vs the power required to operate the IT equipment. EUE, the metric that the EPA is using to award Energy Star to data centers, measures the power all the way through the distribution grid from the generating station. Have a really efficient data center but you’re in a lousy place on the grid or a long way from the generator? You get a lousy EUE.

Both of these metrics are critically incomplete and that makes them ridiculously easy to game. Fire up a bunch of extra servers and just leave them idling and watch PUE and EUE come out great! Any efficiency metric that ignores how much actual work is being done for the energy consumed is worthless. This also leads facilities to erroneously being labeled polluters in the press. Polluting isn’t consuming lots of energy to get lots of work done, it’s consuming lots of energy to get little work done. Sort of like stating how much gas a car uses without saying how many miles it goes on that amount of gas, ridiculous.

According to the article, this software ties the value of the work being done directly to the underlying physical infrastructure supporting it. Finally, the last piece of the puzzle, not only from a power efficiency standpoint, but a business bottom line standpoint as well. Operating the data center inefficiently not only isn’t green, it kills the business economically as well (if you don’t want to go green for the sake of it being the right thing to do, then consider the effect on your wallet!).

The software authors say that “today’s power monitoring products focus only on the physical infrastructure, giving insight into how power is delivered to the data center but not insight into why it is being consumed”. They’re spot on, not only in terms of power monitoring, but the defective metrics data centers are being judged on.

Remember, efficiency is power in vs work out. If you’re not measuring both, you’re not measuring data center efficiency, period.

Vern, SwiftWater Telcom

data center facility engineering

Greening the data center : define “green” please …

Tonight I’ve been reading a commentary about green data centers. While I can’t fault most of the recommendations, I think the definition of green-ness is way off base.

Any measurement that attempts to define a data center as green or not green without taking into account the amount of work achieved by that facility is worthless. I’ve taken the Green Grid’s PUE (power usage effectiveness) and the EPAs EUE (energy usage effectiveness) both to task for this. Now I include blanket negative mentions of a facilities carbon footprint.

The problem is that none of these measurements tell you anything about how efficiently a data center uses the energy it consumes to produce useful work. Stating a facility’s carbon footprint without any reference to the work produced by the facility is dishonest as best. The worst part is, without that final piece of the equation, it’s impossible to use any of these metrics to compare different data centers (not that this stops companies from using these as a marketing tool or potential customers from demanding these numbers).

The fact is, data centers can get great looking PUE numbers, have tiny carbon foot prints, and turn out pathetic amounts of work for the energy they consume. Fill a data center with idle servers and I’ll give you a fantastic PUE with ZERO work being done at all! Consuming energy without producing work is NOT in my definition of being green.

Are these metrics totally useless then? Absolutely not. PUE is useful as an internal measurement of cooling efficiency. Improve cooling efficiency, power required to cool goes down, PUE goes down, and that’s a good thing. Carbon footprint can tell us useful things about the energy sources that supply the data center. Replace traditional power sources with green power such as solar or wind (points for power generation on site!) and your carbon footprint goes down for the same amount of useful work, and that’s not a bad thing either.

When it comes right down to it, the only reasonable measurement of efficiency is energy consumed vs work accomplished. You could even expand this to cover the whole system as energy generation efficiency vs energy delivery efficiency vs energy consumed vs work produced to cover everything from the utility power generator to the output of the data center. Now you have something that can be used to compare different data centers, apples to apples. This is what the EPAs EUE should have been.

Efficient power generation, efficient delivery, efficient use of the power to produce as much useful work as possible. This is that real definition of a green data center.

Vern, SwiftWater Telecom

Green data center metrics : It’s ugly …

I just picked up a tweet with a link to Sustainable IT, Green IT . There was one giant thing that jumped out at me:

McKinsey and Company estimates that less than 1 percent of the energy of a data center is actually used to run specific business transactions. The vast majority (up to 90 percent) of energy is consumed by idle computers and for cooling.

I’ve been stating right along that the hyped metrics, PUE (power usage effectiveness) and the EPAs EUE (energy usage effectiveness) measure entirely the wrong things to be useful. Any efficiency metric that ignores the amount of actual work being accomplished with the power consumed is not only terribly misleading, but is also subject to being manipulated (fire up a pile of idle servers and watch your PUE improve!).

So now we have a crack at a real number. Only 1% of consumed energy is going to real work. Frankly, that’s pathetic. Now you know what energy inefficient servers, underutilized servers, legacy cooling systems and inefficient cooling designs, and legacy AC power and UPS systems with multiple conversions through energy wasting transformers are costing you.

If people won’t go green because it’s the right thing to do, how about the hit to the wallet? It’s like shoveling money down a rat hole.

What are YOU going to do to stop throwing money away?

Vern, SwiftWater Telecom

data center facility engineering

Greening the data center: The EPA gives us another crazy metric.

Today I’ve been readin about the EPA using EUE for Energy Star certification of data centers. Just when I thought these things couldn’t get any nuttier.

EUE in short is the amount of energy supplied at the supplying point vs the amount of energy used by the IT equipment. EUE (energy usage efficiency) is very similar to PUE (power usage efficiency) with two notable exceptions. PUE covers only electrical usage where EUE covers all energy coming in to the data center. PUE also only covers electrical power from the entrance to the facility, EUE covers the energy from the source.

So, how does this impact data centers? First, since this measures usage at the source, the data center gets penalized for losses in the power grid. Be ready to build your next data center right next to the generating station (not to mention any other required utility).

Second, now, all energy supplied is now counted. Using utility supplied water for lossy cooling? Prepare to get whacked for it!

Now, I’m not suggesting this is all a bad thing. This metric will drive the use of green techniques such as free air cooling and self generated power.

The biggest problem with EUE is exactly the same as PUE. The only sane measurement of efficiency is energy in vs work out. Both EUE and PUE totally ignore the amount of work being accomplished with the energy. This produces results that range from simply misleading to totally useless.

Can’ t we do better than this?

Vern, SwiftWater Telecom
data center, web hosting, Internet engineering