H.R. 6, The Energy Independence and Security Act of 2007
by Hu Yoshida on Dec 19, 2007
H.R. 6 was signed into law by President Bush today. The provisions of this act includes raising the fuel efficiency of cars, trucks, and SUVs to 35 mpg by 2020, increase the renewable fuel standard to 36 billion gallons by 2022, and enact a national efficiency standard for light bulbs. While there are some concerns about the cost to consumers and the hazordous wastes that will be generated by renewable energy like ethonol and the disposal of flourecent lighting, it is considered to be a major step forward.
In view of the EPA Energy STAR report on Server and Data Center Energy Efficiency report 109-431 delivered this Fall, I was interested in seeing if there were any implications for data centers in this act.
My review did not find any thing specific to IT. One which will have an effect was the acceleration of energy efficienciency technologies in Federal buildings and the requirement for Federal buildings that are remodeled or newly constructed in 2010 to reduce fossil fuel generated energy consumption by 55% by 2010 and 100% by 2030. The Secretary of Energy is directed to identify a Green Building certification for Federal buildings.
This will apply to Federal data centers which will eventually set the standard for all data centers. One of the things we need to do now is evaluate what we are consuming today in fossil fuel energy consumption.
Comments (2 )
Hu, In all this discussion on data center energy efficiency, I don’t see any discussion on water cooled servers and storage. You and I are both old enough to remember water-cooled mainframes, and you also know that water is substantially more efficient as a heat transfer system than air. I spoke to one super-computing center in Europe last year that hasn’t turned on its HVAC system in 5 years. All the racks are water cooled. What do you think it would take to get water back in commercial data centers cooling those racks and racks of low-cost blade computers?
While I’m not that old, I continue to be amused at how many problems the mainframe community has solved decades before it becomes cutting edge in the open systems space. Sun is going back to the future by heading down the water cooled path (stream?) with their “Black Box” data center – though it isn’t the direct water cooled that was available with the mainframe.
The “buy it in little pieces” nature of Open Systems is going to make reintroducing water back into the data center difficult, because manufactures will have to continue to produce systems that can be air cooled to maintain their economies of scale. It’s only when systems are purchased by the multi-rack full (with volume discounts to boot) do people go “Oh I wish we could do something different on the cooling front.” So what we, both customers and manufactures find ourselves in is a local minimum of mass produced lowest unit cost purchasing, where it’s going to take some energy (read investment) to realize the cooling optimizations that have been available to the industry for decades.
One possible way to move us out of current rut would be to define a “standard” water cooled cabinet. (Standard is in quotes because as we all know there is no such thing as a standard cabinet, just a set of specifications that most rack mounting systems can adapt to.) The idea would be define a water cooled rack, that all vendors could interface into. That way they can innovate in the electronics where they have domain expertise and leverage a common cooling infrastructure where they are not. Customers can deploy the racks and hook up water cooled systems as they become available. Who knows, collectively we might even be able to convince governments to subsidize standard water-cooled rack deployments if they can be shown to increase efficiency and reduce carbon.