In the context of servers, virtualization is a way to address partial load. At its San Jose, California, data center, NTT America deployed a cooling-controls solution that addressed partial load on the cooling system. The solution is by Vigilent, and the companies call this “virtualized cooling.” In essence, the system automatically determines where in the data center cooling is needed and how much, delivers the necessary amount of cooling where it is required and puts unneeded Computer Room Air Conditioning (CRAC) units on standby.

As a result of the deployment, NTT saw its 82,000-sq-ft 8MW data center reduce power consumption and system maintenance costs. It also lowered site Power Usage Effectiveness (PUE) and improved reliability.

Tarif Abboushi, VP of data center infrastructure at NTT America, says the system has made the data center manager’s life easier, especially because the facility has a distributed cooling system – one that uses direct-expansion CRAC units instead of a central plant.

Intelligent controls

Vigilent’s deployment included temperature thermistors, wireless transmitters, CRAC controllers, wireless receivers and a workstation that conducts analytics based on sensor data and controls the cooling system. Two thermistors were deployed on every fifth rack. The sensors take temperature readings in various data center points of interest and send this wirelessly to the server. The server is running Vigilent’s Intelligent Analytics software, which does all the magic, determining where cooling is needed and how much, making dynamic adjustments when necessary.

If the temperature in one particular spot goes above a pre-set threshold, the server sends a wireless signal to a CRAC controller in the appropriate location on the floor, which then ramps up that particular CRAC.

NTT saw a 17% reduction in total energy consumed by the CRACs in San Jose after it deployed the solution in 2011. The data center now has better thermal management capabilities, with higher return-air temperature and reduced temperature variation.

The facility has about 100 CRAC units. By putting unnecessary ones in hot-standby mode, Vigilent increases reliability by making the cooling system more redundant. More CRAC units are available on standby in case any of the active ones fail.

In addition to energy savings and reliability improvements, the system has also become cheaper to maintain. Using fewer CRAC units means there is less wear and tear on the parts, Abboushi explains.

Vigilent’s software also uses historical data to predict what cooling-load adjustments may be necessary in future. Some of the CRAC units at NTT’s data center are 12 years old and have a “relatively high events-incidence rate”, Abboushi says. The Vigilent system helps mitigate possible consequences of such ‘events’ in the time it takes the company to replace the old units.

Lots of unnecessary CRACs

Besides the San Jose data center, NTT deployed the solution at its Sterling, Virginia, facility which is of similar capacity but has a central chilled-water plant and 75 CRAC units. Here it achieved a 35% reduction in energy use. Between the two data centers, deployment of the control system resulted in switching more than 70 CRAC and CRAH (Computer Room Air Handler) units to hot standby. The two data centers’ combined PUE improvements were in the 5-10% range.
NTT estimates that as a result of these improvements, the two data centers will reduce associated emissions of carbon dioxide by nearly 9m pounds.

Abboushi says NTT now plans to deploy Vigilent in other data centers where it controls infrastructure – mainly in facilities not fully loaded, “which is where the solution’s benefits are maximized,” he says.

In addition to Vigilent, NTT uses a number of other Data Center Infrastructure Management (DCIM) tools. Its primary solution is Data Trax..NTT did not examine in depth any other cooling-control solutions before trying out Vigilent and agreeing to buy. “We did not try any others before we implemented the test deployment of Vigilent’s system, the results of which drove our decision to buy,” Abboushi says. It did not take any data center downtime to deploy the Vigilent system, which was key “to the positive experience we have had with the system.”

Here to stay

We asked Abboushi whether he thought dynamic control of cooling systems in the data center was the future, to which he replied: “For NTT, dynamic cooling control in data centers is the present, and based on our experience I do believe it will proliferate in the marketplace. Why would it not, when it offers carbon-footprint reduction, increased capacity and redundancy, reduced maintenance and reduced PUE, all wrapped in an attractive return on investment?”

NTT and Vigilent presented the case study at the Uptime Institute symposium in the Silicon Valley in May.

This article first appeared in FOCUS magazine. To register for FOCUS digital editions, click here.