The Information Technology industry has no shortage of buzzwords and more are created every single day. Some have real business or social benefits and others... well, not so much. One area receiving significant attention these days is the concept of Green IT, or providing computing services in an environmentally conscious way.
Most Green IT methods involve significant restructuring of an organization's server and storage infrastructure and may extend to the desktop level. It doesn't take a whole lot to start making an impact either. I'll discuss some starting points for greening IT. We're doing some of these things at Westminster College to reduce our energy costs and our impact on the environment.
Measuring energy usage
Metrics are important in order to assess the success or failure of any project. If you undertake an initiative to reduce energy costs in your data center, you need to know what kind of power is being consumed by your existing servers. One simple way to accomplish this effort is to install a device between the power source and each server and desktop model and measure actual usage for a period of time. One such device, the Kill A Watt, is very simplistic but works very well.
Record the information you get so that you can use it later on to determine ROI for any projects you may undertake to reduce energy usage.
Begin considering green projects
- With some metrics in hand, you can start considering projects that may have tremendous ROI. I would argue that your ROI will not come from energy savings alone, though. In most cases, the savings generated by undertaking projects designed to save energy will be generated more readily in other areas; energy savings may result but will likely be incremental sums only. The list below is in no way intended to be an exhaustive resource but it does provide some ideas about areas where you may see savings.
- Server consolidation. It's pretty obvious that reducing the amount of hardware in the data center will reduce the amount of electricity consumed in the data center. But consider also ancillary savings. When you have fewer servers running, you also have less heat output which means that your cooling costs go down, too. If your consolidation project is highly efficient, you might even be able to cool a smaller area if your resulting space needs are reduced. Although the direct savings are somewhat difficult to calculate, reducing cooling needs does reduce money spent. A server consolidation project of any reasonable size will probably yield fairly quick ROI; ROI is likely within the replacement cycle for the servers.
- SAN infrastructure. Server consolidation projects are often accompanied by the implementation of a SAN. SANs are not generally implemented with direct cost savings in mind but are instead implemented to provide additional levels of service and availability. However, they can be used to assist in the greening of the IT infrastructure. Consider the process by which many IT organizations provision server storage. Thinking long-term, many people buy more direct-attached storage than they need, guarding against having to add disks later on. A typical 2U server can generally house six disks. Suppose you have a data center with fifty of these 2U servers. That's 300 disks, all requiring electricity to operate. If this imaginary organization is like most, in many of those servers, a lot of this direct-attached storage will go unused. This organization could implement a SAN that consists of thirty or forty disks and see a reduction in energy usage. Each server could then be provisioned with nothing more than a pair of disks for the local OS, moving this data center from 300 disks to, say, 140. Now, this undertaking alone will not likely result in cost savings as the price of the SAN will probably far outweigh the energy savings, but used in conjunction with a server consolidation project, could yield significant gains.
- Terminal-based desktops or VDI initiatives. Desktop computing may undergo something of a regression in the coming years as more and more IT organizations begin to realize the benefits of a centrally maintained desktop infrastructure. I used the word "regression" because that's where we started with this whole computing thing and we're heading back! This isn't a bad thing, either. If you look at an organization that has 100 desktop computers, those desktop computers can probably be supported on three or four servers and the desktops replaced with thin clients. Thin clients use a lot less energy that their fat counterparts and also have the advantage of being connected to a centrally managed server farm which is more easily maintained than 100 separate computers.
- Buy more power efficient desktops and servers. And now for a suggestion that doesn't require a wholesale rethinking of your entire organization! Most organizations have their servers and desktops on some kind of replacement cycle. When the time comes to replace a batch of computers, look at the power supply options that are available from your vendor of choice. Power supplies are becoming more efficient and some vendors make available higher efficiency units. Before you pull the trigger on a barebones system, do a cost/savings analysis on higher efficiency hardware. You may find that, depending on how your systems are used, that there is savings to be had.
Since 1994, Scott Lowe has been providing technology solutions to a variety of organizations. After spending 10 years in multiple CIO roles, Scott is now an independent consultant, blogger, author, owner of The 1610 Group, and a Senior IT Executive with CampusWorks, Inc. Scott is available for consulting, writing, and speaking engagements and can be reached at firstname.lastname@example.org.