It sounds like a simple question but answering “Where do YouTube videos come from, Mommy?” requires studying the millions of files, photos and videos that are stored in giant data centers –warehouses with racks of servers that work around the clock. All those servers require electricity to operate; heating and cooling for those facilities and back-up power in case of emergency.
Sun Microsystems Inc. undertook a 10-year-plan and massive consolidation of its server network and some unique ways to lower costs, reduce pollution and cut electric use. One of the company’s server facilities in Colorado uses water-powered cooling – much like 18th century mills used water to drive mechanical equipment.
A data center built in 2003 used an average of 40-watts-per-square foot, according to Anthony Cataliotti, a manager for Sun’s Data Center Efficiency Consulting. By 2005, that figure had tripled and current facilities are now looking at 800-watts-per-square foot.
“There have been extraordinary leaps in the density of equipment,” he told a recent Sustainable Energy workshop in Boston. “I used to run data centers but I never saw the power bill. Business, IT and facilities people may all have different goals.”
Sun’s plan meant getting different departments on the same page – using business intelligence modeling to create 3-D data-rich coordination and match heating/cooling to IT workload. The company cut power consumption by more than half, consolidating 15 sites into two more-efficient facilities, Cataliotti says.
Only about five percent of U.S. companies have looked at their data and network use as a cost of daily business, according to one Sun project manager. Doing that sort of analysis properly means starting with the right tools to measure all related electricity use and their costs. Visualization tools such as TIBCO Spotfire easily create maps or diagrams of daily or hourly electrical use or demand for resources. Once the data is compiled, accurate projections of future needs can be built. And managers can use “dashboards” to monitor real-time information, allowing them to allocate resources where they are needed.
“We went up a learning curve, but saw an order of magnitude improvement,” Cataliotti says. The result was over $1 million in rebates from the electric utility and a cut in the company’s power rates.
Spotfire Blogging Team