McKinsey: Measure Data Center Efficiency Like Car Fuel Efficiency

Despite the increasing urgency to get a hold of soaring energy demands in IT, costs continue to spiral out of control. Energy costs are increasing by 16% every year, while greenhouse gas emissions from data centers have already passed the output of Argentina and are due to pass those of all airlines by 2020.

But, according to a new report by McKinsey & Company and the Uptime institute, it doesn't have to be that way. The report recommended companies adopt a new metric called Corporate Average Data Efficiency (CADE) that combines both IT and facilities costs to monitor energy use, and create "energy czar" positions to manage energy efficiency.

"We need to know better what's coming into and out of our data centers," William Forrest, associate principal for IT at McKinsey, said in a conference call detailing the report. He added that data center growth is beginning to cause significant strain on IT budgets and will continue to do so as companies add more servers unless they begin to focus heavily on increasing energy efficiency. Companies like Microsoft, Google, Citibank, and AT&T are spending hundreds of millions of dollars on new data centers and are being forced to look for ways to be more energy efficient.

According to the report, there's too little good demand and capacity planning within and across IT, business and facilities management functions. That translates to 6% average server utilization and 56% facility utilization. McKinsey found that in some companies, as many as 30% of servers are just wasting space and energy, but are really dead. Meanwhile, top execs aren't holding CIOs accountable for capital and operational expenditures in the data center because they don't have the data to know any better.

"Most organizations could double their energy efficiency by 2012 if they just tried," Forrest said. "In many instances, boards and even CIOs don't participate in the way they should in how these major expenditures are designed and operated."

The report proposes a three-part solution to reversing the trend. One, companies must improve IT asset management with best practices, virtualization and by setting energy efficiency criteria for hardware procurement. Second, companies need to keep track of total cost of ownership of their data centers. And last, companies need to set and execute plans to increase energy efficiency by setting measureable, achievable goals and appointing "energy czars" to carry out the plans.

New help could come in the way of CADE. "It ends you up with a single thing that says this is the miles per gallon that we're getting out of our data center," Forrest said. "It's a simple standard like the CAFE standards for automobiles."

CADE measures total data center efficiency as a percent. The higher the result, the more energy efficient the data center. It takes into consideration average CPU utilization, total IT load, facility capacity and the total energy consumed by a data center. However, it's still missing a few pieces, such as the average energy efficiency for servers, storage and networking equipment.

"This is the first metric I've seen that combines facilities and IT. Each constituency tends to look at each part of the problem, and when you look at it separately, it doesn't seem too bad," Kenneth Brill, Uptime Institute's founder and executive director, said on the call. "The shocker is when you do put them together, it's clear that people should do something... My sense is that we don't need government regulation, because this is all in self interest."

The report also recommends a series of ten steps companies should take in their data center, including creating an energy efficiency dashboard, sealing cable cutoffs, turning off and removing excess hardware, increasing temperatures, virtualizing, and upgrading equipment.