- Silicon Valley's 19 Coolest Places to Work
- Is Windows 8 Development Worth the Trouble?
- 8 Books Every IT Leader Should Read This Year
- 10 Hot Hadoop Startups to Watch
Industry analysis by expert Joanie Wexler, plus links to the day's wireless news headlines
Like many organizations, the University of California, Berkeley, has been eyeing the latest data center technologies to increase overall energy efficiencies. The university's most recent challenge was to expand server capacity without overloading its existing cooling system or having to add expensive air conditioning capacity.
The university late last year planned to install a high-performance computing cluster for scientific research applications that would add 10 server racks to Berkeley's 10,000 square foot data center. The servers would also increase uninterruptible power supply output by 20% – from 400 kilowatts to 500 kilowatts, says Steve Aguirre, manager of data center and production control services.
"We needed a way to get specific thermal data and to centrally monitor power use on our IT and cooling devices to avoid adding high-heat equipment in areas that had already been saturated," Aguirre says.
So he began testing a wireless sensor-based monitoring system from San Francisco start-up Arch Rock to baseline key metrics. Arch Rock's Energy Optimizer wireless sensor-based system measures temperature, airflow, pressure and other metrics in data centers at every spot where sensors are mounted. In UC-Berkeley's case, the system collects about 160 measurements across various pieces of equipment; top, middle and bottom.
As a result of having real-time, 24x7, visibility into power and thermal conditions, "we're doing 20% more with our existing assets," says Aguirre, who says his plan is to now install the system throughout Berkeley's whole production data center "probably by the end of the calendar year."
Wireless makes it easier to get more granular readings from more places because cabling doesn't have to be run to each sensor. Also, Arch Rock's system uses IP over the air so that the data can be collected and forwarded over the corporate IP network for correlation and display in Arch Rock's Energy Portal Web-based monitoring application.
Arch Rock's Energy Optimizer and similar products from competitors such as SynapSense offer wireless sensors to help ease the monitoring of key metric thresholds in revamped data centers and to correlate power and cooling system efficiencies. Berkeley's Aguirre is using the Arch Rock system to determine how much power is being consumed by the equipment load and how much is consumed by the supporting power equipment – a ratio called power utilization effectiveness (PUE). Using the wireless tool, the university is shooting for a strong PUE of 1.3, Aguirre says.
Why run wireless sensors instead of wired ones?
"I pay for network connections, and it would be cost-prohibitive to pay [the IT department] for those network connections every month," Aguirre explains. "It's more effective to use Arch Rock; the startup costs are about the same [as a wired sensor network] but there are no monthly recurring costs."
On the other hand, some might argue that battery-based sensors will cause issues when the batteries start dying and must be replaced here and there throughout the installation. According to Arch Rock, its Energy Optimizer sensors should last three to five years.
Joanie Wexler is an independent networking technology writer/editor in Silicon Valley.