A group supported by the federal Environmental Protection Agency (EPA) plans to release a second draft of an energy efficiency standard for 1U and 2U rack servers this Friday.
Jonathan Koomey, a professor at Stanford University and staff scientist at the Lawrence Berkeley National Laboratory, said in a presentation at
The group creating the standard includes representatives from major chipmakers and server manufacturers, as well as the Lawrence Berkeley National Lab, Stanford and the Uptime Institute.
The energy standard would measure a server's performance under different workloads compared to how much power the server draws on those jobs. If approved, data center managers conscious of how much power their servers are drawing could use the standard as one factor in determining what systems they want in their facility.
HP tests RFID in asset management
Hewlett-Packard Co. (HP) has tested the use of radio frequency identification (RFID) tags to track data center hardware with 170-store grocery chain Meijer Inc.
HP has been developing RFID technology for about four years, and project manager Cyril Brignone said he doesn't know when the RFID asset tracking will be ready for mass production, but it will likely be in years, not months. The company plans to improve on the technology and have other users test it out. Meijer tested the product for three months.
"It's a tiny little tag that's attached to the server," Brignone said. "More or less, the system can tell you that it's in that rack in that aisle, and it's the first server in the rack. It can tell you exactly where they are in the rack."
With the technology, RFID tags report back to system software indicating where hardware assets, such as servers, are located. If servers are moved between cabinets or within racks, or if a new server is popped into the data center, the software could theoretically track that. HP also wants the software to be able to have a historical record of where certain servers have been and to where they have moved.
IBM develops capillary system for cooling
IBM has developed a capillary system that distributes paste more evenly within a microchip, leading to what it says is a marked improvement in chip cooling.
Processors use a paste to allow chips to expand and shrink as they go through the heating and cooling cycles.
What IBM has developed in its research lab in Zurich, Switzerland, is a different way to distribute that paste, which it says leads to a tenfold improvement in the transfer between the hot spots and the cooling components of the chip. This new system acts in a similar way to how blood is carried through the circulatory system, or how a tree spreads its nutrients from the roots to the branches and leaves.
Big Blue is also working on another chip cooling technology that sprays water directly onto the back of a chip and sucks it back off in a closed system. The company calls it direct jet impingement, and it includes the use of up to 50,000 nozzles that squirt the water on and cool the chip without the liquid hitting the chip's electronics.
With direct jet impingement, IBM says it can cool 370W per square centimeter compared to current air-cooled techniques that cool about 75W per square centimeter.
Liquid-cooling specialists ISR Inc, based in Liberty Lake, Wash., recently rolled out a chip-level cooling product of a similar design.
Let us know what you think about the story; e-mail: Mark Fontecchio, News Writer