This week's article in Network World about green data centers is interesting because it reflects the IT industry's growing awareness that we have a serious social responsibility to take care of the environment. While this sounds positive there's a real risk that things that appear to be "green" will turn out over the long run to be quite the opposite.
In our story you can see that a major focus of the organizations is to reduce electrical power wasted by staff. That's an easy cause for most companies to get behind because not only is it green, it has a really powerful additional benefit: It increases profits. This is a win-win situation; less electricity equals less greenhouse gases and bragging rights and greater profits, which makes shareholders happy.
But what about what is going on in the actual data centers? I recently read "10 Simple Steps to a Green Data Center" published in CIO Magazine just more than a year ago. The steps came from Dave Douglas, vice president of eco-responsibility at Sun, and the first actionable tip was, "If you have a server that's six years old, Moore's Law tells us that the underlying silicon technology will have improved 16 times since you deployed that system. Not all of that 16x may translate into savings in energy or space, but a lot of it will."
Here's the plan: Take a load of old servers and replace them with a few new machines that use virtualization to consolidate the operating systems that ran on the old boxes. Great, so we gain some floor space (not a green issue) and we reduce power consumption, but by how much? Perhaps we replace four 500W power supplies with one 1700W power supply for a 15% saving. On the other hand, it might be only a three for one replacement which could increase power consumption by 13%!
But whether or not we save power, there's still a question of whether the idea of retiring old servers is really a green idea. The issue is that the eco-value of replacing the server isn't just the money saved by using less electricity, it is also about the total environmental cost of building the new machine (the impact of the pollution involved and the cost of its abatement) and the environmental costs of disposing of the old machines.
From the perspective of a business the accounting may make sense, but from an ecological accounting perspective the bottom line could easily be pretty ugly. For example, the cost of dealing with heavy metal pollution from chip fabs is enormous. Fabs in the United States are governed by laws and regulations, but outside of the country, well, not so much, and even if those fabs are 6,000 miles away they can still have an impact in Los Angeles and New York. It is this bigger picture that has to be accounted for before you can say that any given eco-program has a positive benefit.
Let me give you another example: A while ago I talked to the CEO of a hosting company who was pushing the idea that it was green because it purchased Renewable Energy Certificates and had a load of solar panels in the parking lot. When I asked whether the company had analyzed its total "greenness", which would include the environmental costs of the rest of the operations and the construction and disposal of the solar panels, he got annoyed. He seemed to think that wasn't the point, but in reality that is exactly the point.
Until you have done a real accounting job that includes the big picture, you haven't saved the planet. At best, you've just built a green image and saved a few bucks.