Can’t we just let the market handle it?


Well of course not, even when the market is already moving to the desired result. Some elected official has to jump in front of the parade and show “leadership.”

Congress to Vote on Data Center Power Bill

The House of Representatives is expected to vote July 12 on a bill asking that the Environmental Protection Agency conduct a study on power consumption in data centers, what chip makers and systems manufacturers are doing to increase energy efficiency and what incentives could be used to convince people to adopt energy-efficient data center technology.

AMD and Intel are currently haranguing this market with various notions of “computer processing power per watt.” This bill is just a salvo in that marketing war. Private industry could easily come to some agreement about a metric for this, if it mattered, and as the article states, customers are already aware of the issue.

The bill looks at what is becoming a rising concern in the technology industry. Data center operators are seeing energy costs soar due to more power processors, increasing server density and rising power costs. The heat generated from the newer systems also is forcing businesses to spend more to cool the facilities.

…”This is broad-based support for something that really is a no-brainer,” said [Steve Kester, manager of government relations for AMD] …, who was in Washington for the House discussion on the bill. “We have to address this. It’s a critical issue. And the good thing is that, at the end of the day, it’s going to save everyone money, the industry, the government and even the consumer who uses their computer at home.”

Mr. Kester, lobbyist, is right about it being a no-brainer. It is already being done without federal bullshit intervention. Probably this is why a couple of members of the House are grandstanding with it. No matter what, they’ll have a successful outcome. Left to itself, the market will work this out anyway, based on the selfish interests of the buyers of electricity. Instead, we’re paying Congresscritters 133 million dollars (including 50% for benefits, probably quite a low estimate) a year to direct the employees of a federal agency to spend money studying the “problem.” Call it another $200K for the EPA study. Then there’s 5 minutes of the President’s time spent signing the bill and giving pens to the smiling Rep. Anna Eshoo, D-Calif. and Rep. Michael Rogers, R-Mich.

Five minutes is mainly a lost opportunity cost – the President could have been thinking about something important; something, maybe, actually within the Constitutional purview of the federal government. Let’s call that cost $50K. If the President were smarter it might be $100K.

If our assembled representatives spend an average of one hour each on this bill, we are privileged to pay them $90K for their theatrics. Add that to the EPA and Presidential costs and you get $340K for setting up the opportunity to meddle. Staff costs are incalculable.

I will grant that there is a “saved opportunity cost” in having both Houses of Congress discuss this instead of promulgating some new regulation, so let’s round the money they’re pissing away to $300K.

Unfortunately, that will not be the end of it. The regulation and subsidies likely to result from a study required to findwhat incentives could be used to convince people” will cost, at least, millions. Much of it finding its way to AMD and Intel. Beware the Industrial-Environmental-Government Complex, the government participates in that cabal twice.

Do I have a better way? Of course. I offer this formula to the EPA, gratis and off the top of my head, as the standard. It will work as well as what they will come up with in 60 days of study. (Total data center power consumption in milliwatts, divided by (transactions per second times transaction complexity factor times total bytes transferred)) times the reciprocal of Planck’s Constant. I threw that last in just for the scientific gloss. Since this is a federal government standard it is mandatory to ignore actual local electricity costs.

I would be happy to offer my services (for half the estimated EPA study cost) to come up with the definitive formula for transaction complexity. Or, they could use some that already exist.

In lieu of this “Watt?” project, I’d suggest that the EPA be kept busy calculating the “Kyoto shortfall cost” per breath of each member of Congress, in terms of both generation of carbon dioxide and direct hot air emission. What “incentives could be used to convince people” to reduce this? Universal sufferage doesn’t seem to have worked.

I’ll close with another free idea that would save millions: locate major data centers in areas with cold winters. You design the data center buildings to be able recycle the heat from the computers to warm the building. For a good 5 months a year your space heating costs go down in direct proportion to the heat generated by running the data center. This was done, to my direct knowledge, as early as 1973.

Comments