In the latest of our conversations with compliance and governance executives, we catch up with Steve Sams, head of Project Big Green at IBM, Corp. Readers can also visit our archive of Q&A interviews.
Steven L. Sams is vice president of Site and Facilities Services in IBM Global Technology Services. He is responsible for a worldwide organization that is dedicated to helping clients identify their requirements, current capabilities, and best options for data centers and other specialized facilities like green buildings and clean rooms. This includes building new facilities, and optimizing, relocating or consolidating existing facilities. The Site and Facilities Services organization that Steve oversees includes 500 specialists around the globe that have built over 30 million square feet of customer raised floor for clients from Boston to Bangalore.
Prior to this role, Steve was vice president, IBM Corporate Strategy, where he was responsible for the identification and development of key profitable growth opportunities across IBM.
Steven has received numerous IBM awards throughout his career. He has also been quoted and published in numerous publications on a range of topics, most notably on the power and cooling issues facing companies today. He’s recently been quoted in BusinessWeek, Bloomberg and ComputerWorld and has been published in the Harvard Business Review and Financial Times.
In May 2007, IBM Global Technology Services launched “Project Big Green,” committing to make IBM data centers more energy efficient. The cost to do so: an estimated $1 billion. Annually.
But, IBM officials say, what might seem like a hefty price tag is actually quite reasonable in terms of long-term energy-saving costs. The move not only makes environmental sense, but financial sense as well, they argue.
Data centers consume huge amounts of energy—10 to 40 times as much as conventional office space, IBM estimates. Keeping all those blazing computers from melting themselves consumes on average about 45 percent of a data center’s electricity, compared to the 30 percent actually used to keep the machines running, the company says. Project Big Green helps cut that usage by, among other things, consolidating servers and storage, switching to more energy-efficient machines, and using power management more aggressively.
According to Steve Sams, vice president of facilities services for IBM Global Technology Services and the point man for Project Big Green, the effort is booming. Compliance Week caught up with Sams to chat about the project and ways to save money, energy, and the environment with more efficient data centers.
What does “sustainable IT” mean to you?
Sustainable IT is really about providing the lowest-energy-consumption IT environments for a client. Using techniques and technologies available today, we can reduce typical energy consumption in the 40 to 50 percent range. For a typical data center in the United States, which is about 25,000 square feet, that’s a savings of $1.3 million a year just in energy consumption. In terms of carbon emissions, it’s the same as taking 1,300 cars off the road.
A second major benefit is that by reducing the energy consumption in an existing data center, you actually give that data center more capacity. Typically, our clients’ data centers are not running out of floor space; technologies are getting smaller. They’re running out of power and cooling capability.
So by significantly reducing energy consumption, you give an existing data center a much longer life. It can save a huge amount of capital and provide new capacity without having to wait 18 to 24 months for a new data center.
There are up-front costs to becoming more efficient. How long is a typical payback period?
We’re finding consistently that the payback period of things that have to do with energy savings is a lot shorter than anybody imagined. There’s typically a lot of low-hanging fruit, and paybacks are as short as two years. We found one client an opportunity to save 53 percent of energy consumption in their current data center. That was extreme. But we’re doing hundreds of these.
At a 50,000-square-foot site in Lexington, Kentucky, we found a few things with a payback of two years or less. But more importantly, the data center was completely out of capacity. So through efficiency, we gave them roughly 7,500 square feet of additional building capacity. They could go out and get new clients, install them, and tap into the revenue stream to pay for energy-efficiency improvements.
How has IBM practiced what it has preached in energy efficiency?
In 1997, IBM’s own internal IT use—laptops, payroll and benefit systems, building systems—was supported by 270 data centers, 31 networks, and 15,000 different applications we ran for ourselves. Between 1997 and 2007, we completely rationalized those environments. Now we run not on 270 data centers, but on 12. One network had up to 11,500 servers; we reduced that to 1,500 through virtualization. We had 8,500 Unix servers and reduced that to about 1,500, also. In that set of activities, we’ve reduced total operating cost by $1.5 billion a year. A significant portion of that is in energy savings.
How common is such server proliferation?
When I talk to chief information officers of large corporations and ask how many data centers they have, I typically hear “13” or “27” or “45,” or some other big number. I’ll ask them: “Did you have a plan to get to [that big number], or did that just happen?” What they admit is it wasn’t based on any strategy.
We’ve worked with lots of companies. We took the Industrial and Commercial Bank of China from 38 centers down to two. It saved them $130 million a year.
How do you get the C-suite on board with the upfront investments?
C-suites may not be measuring objectives associated with environmental improvement. Data centers by nature are power hogs. When I first started talking to CIOs about the whole energy thing, I had a feeling—just a feeling—that they actually didn’t know. So I asked a bunch of them if they were paying the energy bill for their data centers, and the answer was “no.”
The problem is that companies allocate energy bills on an average square-foot basis. So if you’re an IT executive, the bill is one-thirtieth of what you’re actually using. About five years ago, our CFO saw our energy bill and asked one very simple question: “Who are the users of this?” And a month later, he did three simple things: He didn’t worry about energy use in our offices, he gave the bill for our data center to the CIO, and he gave the bill for manufacturing to the head of manufacturing.
What are the main obstacles standing in the way of boosting data center efficiency?
First, corporations need to really hold somebody accountable. You need to connect the cost and the benefit to the same organization. IT’s going to be a big user, and that’s a good place to start. Second, people just do not have the facts. They don’t know how energy efficient they are, and they don’t know where the waste is or is not. They have no idea how to make a business decision around how to make energy-efficient decisions in their organizations, because they don’t know what’s going on. So we spent a year developing energy-efficient assessment tools we can take into a client’s data center.
What data do you gather?
We spend four to six weeks onsite and create a sort of “miles per gallon” for a data center, called “power usage effectiveness,” or PUE. It’s a widely accepted metric developed by the Department of Energy. It measures the efficiency of how much of energy coming into a data center is used for productive work, as opposed to turning on lights, cooling, et cetera. An efficient organization will see 70 percent of their energy used by servers, storage, and telecom gear. A very inefficient one sees about 25 percent used productively.
We then tell the clients where the inefficiencies in the data center really are and give them a series of specific actions to improve their energy efficiency and tell them how much savings it will deliver. Then make a business case for those actions based on cost. So a CIO can make a decision based on a set of facts—whereas today, it’s: “Maybe I should install solar panels, but I have no idea what would generate the most savings at the right price.”