All fields are required.
Data centers represent facility investments worth millions of dollars, simply in their construction. They also incur much higher costs for operations and maintenance than more traditionally owned real estate. So it's important to calculate the total cost of ownership (TCO) differently. Experts, like Peter Sacco, president of PTS Data Center Solutions, Inc., rely on a formula developed by the Uptime Institute.
The Uptime Institute has a spreadsheet tool that incorporates all components of data center costs, including both capital expenditures and operating expenses. The tool is available from the Uptime Institute’s website free of charge.
"Budgeting for data center physical infrastructure costs is just not covered adequately in the traditional real estate method of assigning these costs on a square foot basis," Sacco explains. An alternate method assigns TCO on a per server basis, but Sacco believes that allocating part of the physical infrastructure on a dollars per kilowatt-hour basis and other infrastructure components on a square footage basis makes most financial sense for TCO.
Typically, power and cooling, UPS, battery backup and so on are calculated on the dollars per kilowatt hour method, while raised flooring, lighting, grounding and fire protection are covered best by the dollars per square foot method. Though there are many variables, Sacco finds TCO often breaks down as 80 percent in dollars per kilowatt-hour, with 20 percent in dollars per square foot.
The capital expenditures for initial construction of a data center typically break down as 30 percent for IT equipment, with the rest for infrastructure. Infrastructure costs include the exterior framing, as well as cooling, air handling, backup power, power distribution and power conditioning as well as architectural and engineering fees, interest during the construction phase, land, fire suppression costs, etc.
Operating expenses include electricity costs for the supporting infrastructure, as well as facilities management staff, maintenance, janitorial and landscaping costs, security, property taxes and so on.
Sacco also notes that TCO needs to take into account the data center's Tier level, as that will dictate the level of redundancy. "Tier IV is fully fault tolerant," he says. “You have not just one set but two sets of Tier II infrastructure to factor into TCO." So the dollars per kilowatt hour will be much higher because such applications can tolerate virtually no downtime. Tier IV applications typically are found in large financial institutions and insurance companies where the criticality outweighs the much higher dollars per kilowatt-hour rates.
More typical data center applications are Tier II, with N+1 redundancies. Even at this level, however, TOC will vary depending on whether they are high-density or low-density applications or some combination of the two as is found in many colocation facilities.
"Three elements can significantly impact TCO: total load, total availability required and deployed density," says Sacco. "Increase any one of these elements and the dollars per kilowatt-hour increase substantially."
That said, factoring capital expenditures and operational expenses for TCO ahead of building or revamping a data center offer many opportunities for potential savings, particularly in energy costs. But he points out that increased efficiencies to lower operation expenses will mean larger capital expenditures.
"Generally, the priority is not reducing electrical consumption," Sacco admits. But by using TCO and other computational fluid dynamics modeling tools, data centers "can achieve some good energy efficiency improvements that should be considered."
Learn more and get product information from Schneider Electric
Get monthly email alerts featuring project advice from the field.
Gartner Names Schneider Electric in its Top 15 2015 European Supply Chain Organizations
Tue, 06 Oct 2015 00:00:00 +0100
Prith Banerjee joins Schneider Electric as Chief Technology Officer
Tue, 22 Sep 2015 00:00:00 +0100
Schneider Electric Joins with HKUST—MIT Research Alliance Consortium to Further IoT Innovation
Mon, 21 Sep 2015 00:00:00 +0100