Daily Archives: August 19, 2008

Green Your Data Centers and Keep More Green In Your Bank Account

Supply and Spend Management Professionals everywhere, take heed, you can save bags upon bags of money by forcing IT to Green your data centers – and please the grippies (green hippies) at the same time as you drastically reduce your energy requirements, and, consequentially, your carbon footprint. That’s right, you can save energy and the environment as a side-effect of implementing IT technology that not only saves you money, but performs better overall! And all you have to do is make sure IT buys the right technologies from the right vendors, and not just the vendors that give them the most free toys or the most free passes to the tech expos. So read this post carefully. The technologies described within will not prevent IT from supporting your current application infrastructure or affect any SLAs you might have – although it might mean that IT has to upgrade their skills and / or get comfortable with new technology. (And if you’re not willing to learn on a daily basis, and you’re in IT, you’re in the wrong profession — and shouldn’t be surprised if you’re told to shape up or ship out.)

As I informed you in a recent post on how IT is our greatest threat to our energy future, your corporate data center is a huge energy hog, and likely sucks up considerably more energy than the rest of your office-based operations combined. Why is this the case? First of all, those big-boxed traditional servers suck up considerably more power than an average workstation, especially since they usually have redundant always-on power supplies, with power requirements in the 700W to 1,500W range being quite common. In other words, each server generally sucks up three times the power of a normal workstation, 24/7, whether it is used or not. But mostly your data centers are huge power hogs because of the huge amount of heat a large number of servers in a compressed space will generate and the simple fact that if the room is not cooled to at least normal room temperature, they’ll overheat, melt, and take not only your hardware investment, but your data (which just might be your most valuable asset) with them. Today, it often costs more to keep your data centers cool than it does to fill them with equipment. If you reference a recent article from Hewlett Packard on Electronics-Cooling.com, and recalculate the 3 year energy cost to acquisition cost with today’s prices (where you can get a fully configured 1U server for (well) under 3,000 and where the average retail price of electricity is now over 11c per kWh and rising fast), you find that, today, the average cost of keeping a server powered and cool is at least 30% more than the server itself!

So what can you do? You can start by using servers with power efficient chips, virtualization, and dynamic CPU allocation to reduce your power requirements. These initiatives can decrease power requirements substantially. Lower power hardware requires less watts to run. Virtualization allows you to reduce the number of servers you require by packing more applications onto fewer servers. And dynamic CPU allocation allows you to automatically shift processes between processors and, during down times, power down one or more cores to reduce energy requirements. This means that your server room can run at full capacity during the normal work day when it is being hit hard by every employee in your operation, but automatically power down by 75% (or more) overnight when almost no one is utilizing your systems and networks.

Then you can reduce your cooling requirements by using rack-based (liquid) cooling that focuses on keeping the servers chilled, and not the entire room they are housed in. Even in a jam-packed server room, your servers aren’t even going to take up a third of the space. This means that not only are you chilling a lot more space than you have to, but you have to chill part of the room cooler than it needs to be as, chances are, the vents won’t be right next to all of the servers. However, a rack based solution that only chills the servers only sucks up the energy needed to chill the servers, and this tends to reduce your power consumption by at least 15% to 20% alone! Furthermore, you can then install a heat exchanger to route the vented heat throughout your office building and use your servers to heat your building in the winter! That’s two hits of energy savings for the price of one!

So where can you get this technology? IBM and Sun are making leaps and bounds in virtualization, IBM and HP are leading the way with rack-based cooling solutions, and all three are leading the way in energy efficiency in their server offerings. (However, it appears that IBM has the best dynamic CPU allocation as it’s virtualization is dynamic and allows you to time-slice processes down to 1/10th of a CPU and use shared pools to insure a sleeping core isn’t powered back up until absolutely needed.) So, contrary to some propoganda, you’re not necessarily restricted to one vendor if you want to save energy, and thus you can use the competition between these vendors to save huge bundles of money.

At this point, you’re probably asking “How do I start?” and “How do I figure out how much money I really could save by helping IT go green?“. The answer is you start with a data center audit that examines your current infrastructure, your current network, and your current user support needs, and comes up with an optimal data center design using best-of-breed technology that will meet your current needs with less energy and less hardware investment requirements than it would cost to upgrade your data center using your current technology and infrastructure design. (And yes, you’ll save money on hardware too since virtualization will let you do more with less!) Then, as you replace and retire existing hardware, you can replace it with the right hardware – and cooling systems, for your needs and watch the savings rack up year over year (especially since today’s virtualization platforms tend to be more extensible and upgradeable in addition to lasting longer).

If you’re looking for someone to call to help you with this audit, I’d recommend you consider NCS Network. The first provider in Canada to offer a green data center solution, they have years of experience in designing, maintaining, and auditing data centers. In addition, they have partnerships and relationships with Sun, IBM, vmware, MiTel, Cisco, Wyse, Citrix, Xen Source, red hat, APC, Novell, and, of course, Microsoft. If you mention that you heard about them on Sourcing Innovation and book an audit before September 30, 2008, they’ll give you 10% off of their standard rates.

In full disclosure, although I do not own any shares in NCS Network, or get any commission for referring you to their services, there is a partnership between NCS Network and the doctor‘s company, whereby the doctor may serve as their Chief Software Architect on a consulting basis when an NCS Network client needs a senior software architect and Emerich Winkler of NCS Network may serve as the doctor‘s Chief Network Architect on a consulting basis when one of the doctor‘s client needs a senior network architect. However, if you hire them, and don’t hire me, I get zip, zero, and zilch and there’s really no financial incentive for the doctor to recommend NCS Network to you.