Greenpeace: Giving stick where it is due Mark Bramfitt, PE offers a riposte to FOCUS' views on the Greenpeace report into datacenter powerFocus has come to the datacenter industry’s defense with regard to the recent Greenpeace report that ranks companies on several aspects of their commitment to environmental sustainability. To one extent or another, I disagree with the criticisms, though I do have one or two of my own to add to the list.
But more importantly, the core message of the report stands: where you site your datacenter is almost manifest destiny when it comes to the carbon impact of the power you purchase from the local utility.
Some companies are committed to purchasing renewable power, but many more are seeking low cost electricity, which means they buy the dirtiest power. These operators should take stick for that.
Is Energy Efficiency A Compelling Defense?
The loudest defense offered to counter the criticisms of the industry that the report outlines is that energy efficiency has become a widely adopted practice, and that the report fails to take this into account.
All of the datacenter operators listed in the report should indeed be commended for developing new facilities that incorporate best practice designs for energy efficiency. Many of these new datacenters are pushing the theoretical boundaries of efficiency for cooling and power delivery and conditioning systems, with IT efficiency remaining as the final frontier.
But Greenpeace expressly acknowledged that the companies are doing quite well at energy efficiency: their rating for energy efficiency attempts to acknowledge those players who are both doing a good job and sharing their practices.
The executive summary of the report notes that “There are increasing positive signs of collaboration and open source sharing of best practices…” This is a big change from just a few years ago, when notable players kept their efficiency cards close to their chest, presumably to maintain a competitive advantage.
But more to the point, the adoption of energy efficiency best practices has little to do with siting decisions and power purchase arrangements. I can have the most energy efficient datacenter in the world, but serve it with high carbon content electricity because I built it in the wrong place, or simply want low power costs.
So Greenpeace wasn’t giving the industry “stick” for a lack of focus on energy efficiency – in fact they acknowledge that this is a bright spot. They are criticizing siting and power purchase decisions, which simply cannot be brushed aside with efficiency claims.
Greenpeace’s inability to accurately assess the power draw at the major datacenters of the assessed companies, and the assumption of power generation profiles for each site, are cited as a “suspect approach” that invalidates “any ranking of the companies”.
The report goes to great lengths to explain the ranking methodology for the “Clean Energy Index”, acknowledging that estimating datacenter loads is particularly problematic given the industry’s reticence to release data.
The industry can be as defensive as they want on this score, but Greenpeace has not only described in detail their assumptions and methodology for assessing loads, but has given every company a chance to “come clean” with actual data.
On the other hand, the methodology for determining the power generation profile for each site is unassailable in my view, absent any disclosure from firms who have renewable power purchase agreements or who have onsite generation. (And really, wouldn’t these companies want to disclose that they were doing so?)
Greenpeace uses actual utility generation profiles when it can, which is the best means of assessment available. I have called on the utility industry to provide time-of-day carbon intensity information, which would lower the carbon assessment for high load factor users like datacenters, but until that day comes, the approach used in the report is sufficient.
Finally, even if you don’t support the technical analysis and lay all of the responsibility off on Greenpeace rather than the lack of transparency of the companies, that only calls into question the Clean Energy Index, not the four other ratings presented in the report.
And What Of The Protagonists?
A fair criticism of the report is the list of companies that were assessed – as Ambrose quite wittingly notes, an apples to apples, or rather an Apples to Amazons issue.
I agree that as many as half of the ranked companies aren’t truly in the cloud services game (Twitter? - no way!), and I think Greenpeace maybe intended to help cloud service customers assess the environmental qualifications of potential providers.
But this criticism should be leveled on the basis of that criterion, rather than on the relative sizes of the company’s operations. If you agree that file services companies are cloud service providers, wouldn’t you want to see them ranked even though they don’t compare to the behemoths in the industry?
What Would Make A Fairer Ranking?
As I posited in my initial reaction the report, I find fault with Greenpeace for focusing on coal and nuclear generation as bogies. A better approach would be to assess the amount of generation from each source in the mix, assigning the appropriate carbon intensity to each.
Note that this approach awards nuclear a zero carbon score, while properly assessing some types of renewable generation (biomass and biogas both have associated carbon emissions). The Greenpeace approach doesn’t withstand technical or public policy rigor in my view.
I join Ambrose in thinking that the listed customers are not all of a piece, though in removing some I would seek to add others who are emerging in the cloud services market.
And I’d ask the industry for more transparency: on loads, renewable power purchases, and on their IT efficiency efforts and utilization rates. And if that’s a bridge too far for competitive reasons, then I would ask that the industry develop a suitable framework to make the ratings reasonably insightful and useful to potential cloud service clients.
The Greenpeace report makes several points very well, and where the analysis and assessments have been difficult or are less than ideal, the authors have been forthright in acknowledging gaps.
Many datacenter industry players do indeed deserve carrots for the focus on efficiency in recent years, but many also deserve “stick” for siting decisions, a lack of transparency, and a lack of commitment to responsible, sustainable power purchasing practices.