Five Steps to an Energy Efficient IT Environment

Starting small can provide large benefits
Written by David Chernicoff, Contributor

Earlier this week I received an email from the PR firm for the Green Grid with this intriguing title. I know that there is an innate urge in many people to build and use lists to accomplish tasks, but it certainly seems that there should be more than five steps to making IT greener. And looking at the list (which I will include at the end of this piece), it was definitely more of a generic overview of steps that can be taken to improve your IT efficiency.  But for the holiday season I think steps that involve major policy changes, or revisiting how you manage your business workflow and expenditures would simply be adding major projects to your life, so instead I think I’ll first give you my “Three Simple Steps to Improve the Energy Efficiency of IT.”

  1. Put all the pieces back

I can’t tell you how many times I’ve walked through client or vendor datacenters where there was clear evidence of projects that were in progress or completed and not cleaned up.  When you’ve removed blocking panels from racks, opened up ceiling and floor tiles to move cables or equipment, removed or added systems to racks, or made changes that otherwise affect the way the air flows in your facility don’t forget to put everything back in place.  Close up the tiles, replace panels in the rack, and close up areas that should be closed and open areas that should be open.  When you mess with the airflow in your datacenter I can guarantee that the end results will be a higher expenditure to keep the facility at the proper temperature.  There’s a reason it was put together in that nice, tidy fashion.  Don’t let temporary changes stretch out to the point where the operational costs grow.

2. Turn stuff off

It’s great that you are inventorying the equipment in your racks and are identifying equipment not currently in use.  Unless you’re planning on provisioning unused servers and such pretty soon, just turn them off.  There seems to be a certain amount of inertia in this with lots of shops I’ve been in setting up racks turning everything on, making sure it works, and just leaving it on, even if it isn’t going to be used for a few days. When those days stretch out to weeks or even months it’s just a complete waste of power. That Post-It note stuck on an unused server isn’t saving power.

3. Consolidate

This one is a bit more time consuming, but the resulting savings can be significant. Given the capabilities of network management with virtual ports, routing, etc., there is rarely a good reason to have lots of racks that aren’t completely populated. Unless you have a specific need to run any number of half empty racks spread throughout your datacenter, consolidating the physical servers into as few racks as possible will minimize the area that you need to maintain environmentally correct for the hardware.  Doing this also allows you to take advantage of inexpensive methods of hot-aisle and cold-aisle containment.

That’s it. No rocket science here but rather reminders of some of the simple things that can be done to reduce the operational costs of your datacenters.  And realistically, these simply boil down to good work habits and a little bit of foresight when making changes in your datacenter. Now for those of you who want the higher level view that the Green Grid is providing in this case, here are their steps:

1.            Calculate Your PUE.

The first step in understanding your data energy consumption is to take an initial assessment of the power you’re using in your data centers, or your power usage effectiveness (PUE). This will give you an idea of how efficient your data center is using and sending power to its servers. Typical numbers for people who have never measured before average between 2 and 3 watts of consumption to deliver 1 watt to the servers.

2.            Keep It Cool.

Once you know your PUE, you can start addressing facilities issues, and more specifically, cooling. In a case study published by The Green Grid, 5 cooling initiatives were tested over a period of 8 months. The testing determined overall energy savings to be worth over $300,000 (USD) per year, paying back investments in 15 months. The specific initiatives included installing OEM VSDs, upgrading CRAH units, implementing rack airflow management, moving the CRAH control from the return air temperature to the rack inlet, and increasing the temperature set points of the CRAH and chiller units. Overall, improving cooling technologies can lead to massive energy and cost savings.

3.            Virtualization.

Making better use of your IT resources is another important factor in the energy-saving equation. To make IT machines work more efficiently, virtualization is a key consideration. Steps include taking an inventory of all IT machines operating in your environment and constructing a plan for consolidation based on your needs. After all, every watt of power that a workspace doesn’t need is a watt of power that doesn’t need to be cooled.

4.            The Powers That Be.

Natively controlling the power consumption of IT assets using IP-enabled energy management tools provides major efficiency gains.  Enterprise energy management systems can monitor and control compute, network and storage assets.  Servers are a good place to start and server power management (SPM) is a tool that many data centers are adopting, and can save as much as 33% of the energy used by a server, without effecting performance. There are a variety of ways that data centers of all sizes can take advantage of these new methods. The Green Grid has members-only resources to help understand and tackle the issues around SPM.

5.            Future Forward: Carbon and Water.

Facebook recently became the first company to publicly announce its WUE (water usage effectiveness), a metric developed by The Green Grid. Understanding the WUE – and even CUE for carbon usage effectiveness – measurements will keep you ahead of the game, and with companies like Facebook taking the lead, WUE promises to be an increasingly important measurement going forward.

You’ll see that the methodologies espoused here all are based on the Green Grids measurement and modeling techniques, many of which I am in favor of.  But don’t forget the immediate, and simple, things that can be done to control energy expenditure in your datacenter.


Editorial standards