Green IT: Myths vs. Realities
I'm reading an interesting book by StorageIO Group analyst Greg Schulz titled, "The Green and Virtual Data Center." It got me thinking again about green IT. A few months ago I ranted about the hype aspect of green IT every company jumping on the bandwagon and repositioning products as green. It was very similar to the Information Lifecycle Management (ILM) propaganda that pervaded the storage world a few years ago that lacked much in the way of real substance. It's time to replace green posturing with something more tangible than well-crafted sound bites. To do that, we must look beyond the hype and examine the facts.
This new read from Schulz turned my attention once again to green IT. Perhaps there is hope that the green posturing will be replaced by something more tangible than well-crafted sound bites. For that to happen, though, the myth will have to be separated from the reality.
For example, the book goes over the well-quoted "fact" that servers consume the bulk of electric power in data centers. This is one of those partial truths that is really a distortion. For computer-intensive environments, servers may consume more than 50 percent of electricity if you include the power and cooling used to keep those servers running. If you want to finger one area as consuming 50 percent of total electrical draw, a better contender is cooling.
Right along with the immense amount of PR currently unfolding on the subject of carbon footprints, sustainability and CO2 emissions, a data point is quietly slid in that data centers account for 2 percent of all global CO2 emissions. This is an exaggeration borne of creative accounting. Schulz' numbers reveal that 2 percent of electrical power in the United States is used in data centers (and the number is far less in the rest of the world). Further, in states like California, vehicles are responsible for more CO2 emissions than power generation. So this is just plain false.
Certainly, the United States must reduce the amount of pollution it generates. Certainly, most organizations must find ways to cut costs and power and cooling are prime candidates. But there is no need to bend the truth to make things appear worse than they are or make the numbers scarier than they need to be.
Not surprisingly, some server vendors have taken up environmental scaremongering in an attempt to use fear to get companies to change out their existing inefficient servers for more modern greener models. That's about the same argument as saying anyone with a car older than a year or two must replace it with a Prius or they are committing a crime against our dear old planet. If you really do the math in terms of environmental impact and economics, it actually is far less damaging for me to hold onto my 10 year old Ford Explorer than it would be to purchase a Prius based on the miles I do each year.
It is the same with servers. Why ditch old models that are performing well only to spend plenty to upgrade to the latest and greatest? By all means, refresh your servers when you are good and ready. But watch that you aren't railroaded into it by imaginary math and some fancy footwork about carbon emissions. As the urge in the current climate is to cut back, it is natural that server marketers will step up their efforts and fudge the numbers to align their sales objectives to the ongoing green campaigning. But don't believe the figures you are fed. Check them out in your own environment. Verify everything.
Going green, then, is not something you do in one fell swoop. Do it gradually. Phase out old servers and bring in more efficient models in a way that makes sense technically and economically. Don't be bullied by the hype. Instead, trust your own numbers, and check out any new wonder-servers thoroughly in your own test environment before deciding to change out your entire facility or invest in a completely new power or cooling infrastructure. The danger is that it might cost you a great deal of cash and then deliver only moderate gain.
I'll end with a case in point. A New York City data center is spending more than $5 million installing solar panels due to the green benefits. Senior management bought into this, yet the cost savings are projected to be about $40,000 or so a month on power. Therefore, it will take a decade or more for pay back if the projections are to be believed and if costs don't spiral northwards. So is it the right thing to do NOW to install solar panels? I'll bet there is a better way to green this data center at lower cost and for more immediate return.
In all likelihood, if the organization planned it properly, the right time to install solar panels is probably as part of the establishment of a new highly energy efficient data center project rather than cobbling them on to an old inefficient structure.