Argonne National Laboratory's energy-efficient Blue Gene/P supercomputer.
Source: Argonne National Laboratory
I wrote back in January that data centers cannot be truly green unless operators look past energy efficiency and start looking at clean energy. So, I was pleasantly surprised by Greenpeace’s recent report making largely the same observation. I’m not even bitter that Greenpeace has more drawing power than me (the response to this report alone must have had a measurable carbon output); I’m just happy so many are at least thinking about this issue, even if I tend to disagree with their assessments.
Despite strong arguments against Greenpeace’s gloom-and-doom vision of cloud computing (including from GigaOM Pro’s own Pedro Hernandez), the reality is that greener is not good enough. Yes, data centers are getting more efficient and increased digital media consumption means less manufacturing and deforestation (in the case of e-books), but mere mitigation cannot be the goal. Why? Because if cloud-computing demand outpaces improvements in energy efficiency, we’re still looking at a net positive in terms of data center emissions.
Even if cloud data centers replace inefficient corporate computing with more-efficient processes, we’re still left with skyrocketing demand, from both companies and consumers, for entirely new uses that have no real legacy analogues. Robbing Peter to pay Paul is not the answer to fighting climate change, especially if we still can’t pay Paul.
However, I don’t believe we can rely on cloud providers, or any large data center operators, to make a major green shift without incentivizing them to do so. ZDNet’s Larry Dignan wrote this week that webscale operators will ensure the cloud is green because lower energy bills are good for their bottom lines. This much is true, but cheap energy also is good for their bottom lines, and it doesn’t get much cheaper than coal. Probably more so than with traditional companies, I believe that companies like Google and Facebook really do try to be as green as possible, but economics dictate that they can’t break the bank to achieve that goal. We should laud their hydro-powered data centers while recognizing that they’re still the minority.
That’s why I place the burden squarely on the shoulders of both the state and federal governments. As I wrote in January, tax credits and subsidies will help clean-energy-powered data center projects get off the ground now, and will pay back later when data centers are even bigger business. Another option is to pressure TARP-recipient banks into providing back-loaded loans that accomplish the same thing. (As a resident of Nevada, I happen to know of a place full of free sunlight and 24-hour-a-day personnel, if only solar infrastructure weren’t so darn expensive …)
Change, in this case, has to come from above because, despite Nick Carr’s suggestion that we consider our personal responsibility for this problem, consumers aren’t going to turn off their laptops, cell phones, e-readers and tablets. Nor are most businesses going to install solar arrays that take 20 years to pay back, or buy clean energy at rates far higher than what standard utilities charge. Unfortunately, that’s just too much to ask.
IT is getting greener, and the technologies underlying cloud computing have a lot to do with that. Virtualization, automation, power-management software, lower-power processors – they all help make computing cleaner than it ever has been. But growing demand means these fixes will only do the job for so long before our digital consumption envelops them. It can’t happen overnight, but there’s no harm in starting the migration to truly green IT now.