Tier Ratings, "9's" Help Show Needed Level of Power Reliability

By Rita Tatum  
OTHER PARTS OF THIS ARTICLEPt. 1: Don't Wait for an Emergency to Review Backup Power PlansPt. 2: This PagePt. 3: Power Reliability: Don't Overlook the Importance of OperationsPt. 4: Power Reliability Product Showcase

Discussions about the appropriate target for the backup power system often lead to considerations of "the 9s."

"One 9 means that acceptable power should be available 90 percent of the time," says Johnston. "Two 9s means 99 percent of the time; 99.9 percent is three nines. At five 9s availability (99.999 percent) a mission critical facility will lose power for 5.5 minutes annually."

While that seems an incredibly small amount of time to be down, Johnston notes that lost data is not so quickly recovered. "That 5.5 minute power outage might take a week for the data center to be fully up and running. Even a one-second power outage still could take the data center up to a week to recover, because you have to bring data centers back up slowly," he says.

"Nines equal the number of hours of downtime that is acceptable for operations," says Schlattman. While higher Tier levels sound good when it comes to availability, the cost to achieve that level of performance must also be considered. Schlattman points out that not only does the backup power supply need to be totally redundant at five 9s, but the mechanical systems also require dual paths and full automation. Building automation systems also must be more robust, as the elements they must operate and control are increased at Tier 4.

As a result, companies are looking at lower Tier levels, when feasible, and even combination Tiers. "We find companies that want Tier 3 for essential elements like power supply, but are willing to have Tier 2 for the mechanical elements," Schlattman says.

Another way to address concerns about the cost of new data center space is scalability. Scalable mission critical facilities have infrastructure in place so that they can be expanded as the need grows, rather than being built out completely on day one.

The upfront infrastructure should have enough access to power to support future needs. For example, Data Realty Northern Indiana is a Tier 3 data center that began construction in South Bend, Ind., in January 2012. The 50,000 square feet facility has two substation feeds, each with multiple 138 kVA lines. This allows the developer virtually unlimited power capacity with redundant generators and backup support for 200 watt per square foot configurations. The building is designed in a modular format so that it can be expanded as needs increase.

According to Data Realty Northern Indiana's president, Rich Carlton, the new data center's infrastructure will "match costs to what the computing needs are and thereby be a more scalable model, not only now but also in the future."

Schlattman believes scalability is very important, particularly in infrastructure and backup power. "Typically, the upfront infrastructure may use a bus system to support, for example, 5,000 amperes. In reality, the data center has two power systems, one that is currently operating and one that is offline, so that the load can be transferred as needed," says Schlattman.

Under the bus system, the dual power systems are designed with scalability in mind. The result may be a 50,000 square foot facility that will operate 10,000 square feet on day one.

Using this approach, additional UPS and generators are added as the data center's needs dictate. "You cannot switch out the bus or utility supply," says Schlattman. "But under that infrastructure, you can build a data center that is reliable, scalable and available for some time."

Contact FacilitiesNet Editorial Staff »

  posted on 3/20/2012   Article Use Policy

Related Topics: