Green Data Center Best Practices from Deutsche Bank

Green data centers ensure that every watt IT consumes delivers the greatest business benefit. An expert shares best practices for getting there.

May 17, 2010

By Dave Trowbridge

One of the world's leading global investment banks and provider of financial services, Deutsche Bank, puts a strong emphasis on environmental sustainability with a commitment to make its business operations entirely carbon-neutral from 2013 onwards.

One of the most visible examples of the bank's efforts is its real-time Carbon Counter in New York City's Madison Square Garden. This is a 69-foot billboard that displays a running total of long-lived greenhouse gases in the earth's atmosphere, based on measurements developed by scientists at the Massachusetts Institute of Technology (MIT).

Of course, any corporate green strategy such as Deutsche Bank's involves a lot of behind-the-scenes activity as well. Not surprisingly, much of this is IT-related. News@Cisco recently spoke to Andrew Stokes, Chief Infrastructure Architect at Deutsche Bank. In keeping with the bank's intention to be a "climate ambassador," he was happy to share some of Deutsche Bank's thinking and best practices for green data center design based on the company's experiences.

Q: How does IT fit into Deutsche Bank's overall sustainability efforts?

Deutsche Bank (DB) has three focus areas for its sustainability efforts. One – as a Financial Intermediary, creating emissions trading markets and investing in clean technologies. Two – as an Eco Efficiency Manager – embedding our principles into our daily operations and key initiatives comprising green building & IT operations and switching to renewable energies. Three – as a Climate Ambassador, communicating our sustainability message with clients, policy-makers, our internal staff and the general public through research notes, web sites and public events.

Deutsche Bank IT is a key partner in each of these areas, supporting our business trading and investment activities, engaging in technical dialogue with our vendors and policy-makers (e.g. the US EPA), and driving what we call the "Eight Commitments to Eco-Efficient IT."

Our first commitment is to neutralise the carbon footprint of IT used at DB managed locations in line with the overall DB Group carbon neutrality commitment. We also are focused on our supply chain environmental responsibilities (both for procurement and recycling); using technology to reduce carbon emissions from travel, and reducing our consumption of paper. Other commitments involve energy efficiency, making sure every watt we consume delivers the greatest benefit possible to the business. For example, we're aiming to quadruple the energy efficiency of our major corporate data centers, measured in terms of compute "resource units" per watt, and then doubling our utilization to multiply the benefits.
In the end, that means: how much business work do I get out of each watt consumed? Naturally, this requires a new way of thinking about data centers.

Q: How does this new approach differ from the way you did it before?

The traditional approach has been to buy a data center, gradually fill it up with IT technology, and 10 to 15 years later when that data center gets full, go buy another one.

Technology trends over the past decade have led to a greater demand for power and cooling. Increasing power density requirements in smaller form-factors such as blades, coupled with the increasing need for large computation grids to solve complex business calculations, has really put pressure on the physical data centers. In some cases, those data centers are running out of power and cooling now faster than running out of space. And as a general message, the lifetime of traditionally designed data centers is falling below 10 years.

That's not a sustainable model. We have realized that we need to start thinking about data centers as dynamic eco-systems in which we can stay for much longer periods. This requires a different approach to the past – one where we make innovative steps to increase the efficiency of the data center, whilst challenging ourselves to conserve IT resources where practical.

Q: What role does the network play in a green data center effort, and how much room is there for improvement?

Here, too, we have seen the need to change our thinking and focus more on utilization than simply on connectivity. When you do this, you find there's lots of room for improvement.

Think about a typical data center with tens of thousands of cables or even more, each able to deliver at least 100 megabits of potential bandwidth or much greater. The aggregate utilization of all those cables at any one point is probably well below 1 percent.

Virtualization helps by concentrating business workloads onto fewer physical servers, thereby allowing us to deploy faster network connections that will be better utilized. A well-designed network architecture is of course critical to enable this. With fewer physical cables and ports, we are finding that we are able to increase network energy-efficiency in our facilities, whilst also simplifying the physical topology and improving the resilience.

Q: What would you say is the No. 1 factor in green data center design?

It doesn't matter whether you're building from scratch or upgrading an existing site, understanding your consumption of energy is critical. You don't realize how much room you have for improvement—how much capacity you actually have in your data center—until you really measure what's going on.

For instance, server utilization is a common metric, and the industry average is reported variously between 5 percent and 20 percent. Ours is higher than this, but not by much. (In fact, for several types of our workloads, we intentionally run lower utilization levels in order to achieve the required level of response-time.) Our ambition is to double our overall utilization, but that's only part of one of the three factors that govern the effective IT capability per watt of electricity.

We calculate IT eco-efficiency using three percentages. Data center infrastructure efficiency tells you what percentage of the energy consumed in a data center actually gets to the hardware. Hardware power relative efficiency tells you how much computing capability you are getting from each watt relative to best-in-class hardware; we had to develop our own metric here based on external benchmarks and our selection of a reference server to make a notional 100 percent reference point. And, of course, hardware utilization efficiency gets you more useful work from the available compute cycles. The nice thing is that you can multiply these three percentages to arrive at an overall energy efficiency metric for each facility.

(You can find a more detailed exploration of Deutsche Bank's IT eco-efficiency equation and how it plays out with a given application mix on the Cisco Data Center blog.)

Q: What did you find when you ran the numbers on your Deutsche Bank data centers?

 It was actually quite surprising. Bringing these aspects of efficiency into a single percentage really allows everybody to focus on the size of the opportunity in front of us. The actual metric for us comes out at somewhere less than 10 percent, but the absolute value itself is not really of too much significance, because there are many reasons why it would be impossible to achieve 100 percent. For us, the important message is to look at the directional trend of the metric, to commit ourselves to actions that drive the trend in the right direction. We know that this is a multi-year journey, but we see possibilities for a 4x to 8x improvement if we remain focused for a whole technology refresh cycle.

Q: Finally, where are the biggest opportunities for eco-efficiency in data centers?

You'll find them in both existing and future deployments. We have Deutsche Bank data centers of all ages from almost brand new to some dating back to the 1980s. As we focus on our three aspects of efficiency (i.e. of the facility, the hardware and the utilization), we are continually finding new opportunities for optimization.

Of course, it's in building new facilities that you get to take giant steps, but we are also finding ways to modularize this thinking and apply it back to our older facilities as well. This is probably the biggest opportunity: Taking new concepts and applying them back into existing buildings. For example within our Data Center Economization initiative a lot of small things have been done that are collectively saving thousands of tons of CO2 emissions and leading to significant cost savings, e.g. through sealing up the cable holes in the raised floor, blanking all the spaces in the server racks, reducing air-conditioning costs by raising the temperatures and by broadening the humidity ranges. In some cases, we have even been able to place some of our air conditioning units into standby, as a result of achieving significant energy and air-flow economizations through these activities.

Based on the concepts discussed here, we are bringing in new in-house tools to enable near-real time views of our data center utilization and efficiency. These new ways of viewing the capacity and utilization of our data centers and IT infrastructure will help us to continue driving our transformation objectives.

The best news is, the more you focus on energy efficiency, the more business capability you create from the same power that your facilities use today. It's just a change of thinking required to refocus investments from new data center projects towards sustainable data center projects. Or as we say: "ANDC," meaning "Avoid New Data Centers"!

Share this article: