Mark Bramfitt - Utilities & IT

Reading Now: Hot by Mark Hertsgaard

I can't recommend this work highly enough, because Mark Hertsgaard focuses not solely on climate change mitigation efforts, but more importantly on the adaptation that will be required regardless of the success or failure of mitigation strategies.

I've read abundant commentary on how we must reduce the carbon output of our societies, and of course am a huge proponent of energy efficiency improvements as one slice of that puzzle.

But what struck me is that even if we successfully reduce our carbon emissions to sustainable levels tomorrow, we still face climate change that is already "locked in" from previous activities. 

That climate change will include no small measure of sea level rise (at least two feet based on conservative estimates, and perhaps much more), weather disruptions (mega-storms and changes in rainfall patterns), and potential water and food supply crises.

So we have two immense challenges, not one, and the alternative is "suffering" in Hertsgaard's calculus. We are going to loose land, more people are likely to go hungry, and mass migration will occur - some of this is going to happen even if we are nominally successful at mitigation and adaptation.

For my own part, the book reminded me to double-down on my commitment to finding sustainable solutions for water supply in my community, including the conservation, recharge, and reuse measures we've been considering for some time now.  I'm also meeting this week with our Resource Conservation District to better understand how the rise in San Francisco Bay levels will affect the southern shore of the Sonoma Valley, and what we plan to do to prepare for that.

In summary, Hot is a thought-provoking work, using compelling examples of climate change adaptation strategies from around the globe. It's well worth adding to your reading list.

 

Appeared At: Uptime Forum 2011

Bit of a conference whirlwind in the past few weeks - this is the first of a series recalling memorable themes form the events.

I attended all three days of the 2011 Uptime Forum in Santa Clara two weeks ago (and that's a long slog!), with the added bonus of sheparding two of my utility clients around the conference as they attended for the first time.

The 451 Group and their varied consulting groups (including Uptime) established a broad conference theme around the industry disruption that is or will occur as users move IT activities to cloud providers. I was most interested in the program on day three, which explored the energy and environmental impacts of this trend.

My "star turn" was to act as the moderator at an executive lunch session sponsored by Panduit and Power Analytics (formerly EDSA). They were joined by Byron Washom from the University of California San Diego, who is running what is essentially a highly integrated utility on campus featuring self generation of a variety of types, along with demand control and extensive metering and monitoring.

That setup lead us to our primary discussion, which centered on why data center operators are loathe to consider self generation, demand response, and load shifting opportunities. I had to get things rolling by calling on some participants (it's a good thing I knew at least half of the folks in the room!).

Many commenters posited that critical facility operators are too concerned about reliability to be open to supply and demand management. Byron truly put that in perspective by pointing out that his campus supports data center infrastructure, but that isn't even his critical load - he has medical facilities including a hospital with critical care units, and a storage facility for ice core samples dating back a century.

My ultimate assessment is that data center operators should be thinking about supply and demand management, and that there are good tools (including gear from Panduit and Power Analytics) that can get them there.  The industry is no better or no worse than other sectors when it comes to engaging more fully in these areas. Ultimately, some leaders will break out front, and they're likely to leverage the benefits by being a more competitive cloud service provider, search provider, or e-tailer.

Natural Gas Perhaps Not As Clean As Advertised

The New York Times has an article today reporting the results of a study by an acknowledged opponent of shale gas extraction, noting that methane released during production and distribution may obviate to some degree it's advantage as a cleaner carbon fuel than coal.

The point is that methane is a potent contributor to climate change: it dissipates more quickly into the atmosphere and traps more heat than carbon dioxide, so leaks are damaging. (No one is disputing that natural gas is at least twice as clean as coal when it is burned.)

The natural gas industry of course argues with the estimates of fuel losses throughout the production and delivery system, and whether the losses are higher for shale extraction compared to traditional wells. The utility industry ought to be able to state with a large degree of accuracy losses in distribution. (I know we accurately account for leakage at the water district where I serve on the board.)

This isn't a tempest in a tea cup, but it does point out that all types of energy production and delivery should be evaluated on an end-to-end supply chain perspective. All fuels require energy for processing and delivery (in some parts of the world natural gas is simply flared off as a waste product because the LNG market can't handle the load), and environmental impacts should properly be accounted for.

The issue raised by the report reminds me of the perception of utilities that I found in the information technology industry when I first started getting engaged on energy efficiency. Smart, bright engineers were saying that utilities lost as much as seventy percent of power generation to transmission and distribution losses.

Simply not true. Utilities lose as much as seventy percent of the energy from source fuel (for coal, natural gas, and biomass) in the power plant, but only four to seven percent in line losses. There is also a small band of over-production to provide gris stability (typically three to five percent), but being very clear about where these losses occur is crucial to understanding efficiency and environmental opportunities for improvement.

Slate Covers Greenpeace/Facebook Conflict

Elizabeth Wiengarten at Slate has a great piece on the Greenpeace "feud" with Facebook regarding their selection of a data center site in Oregon that will be served by utility PacifiCorp.

There are a couple of points that deserve amplification: Facebook is not the biggest operator of data centers - not by far. In fact, the new facility (which already has an expansion in the offing), is Facebook's first stand-alone. They are well known as renters of co-location space, but even considering that load, I wouldn't think they're even in the top ten of data center operators in the US.

Next, Facebook and other operators can locate in areas where utilities offer a cleaner generation mix. Claiming that network connectivity and availability of employees are deciding factors in the siting of data centers is hogwash - power costs and availability trump all other considerations, every time, with the exception of financial services companies who crave proximity to trading centers.

(Is Facebook claiming a pool of qualified employees in Prineville? Really? And superior network connectivity?)

So while I too wonder whether Greenpeace's motivations are pure (they could after all go after other high profile targets, or the industry as a whole), they have a valid point. The operators of utility-scale data centers have an obligation to operate as efficiently as possible, and to drive utilities to cleaner generation portfolios by voting with their site selection.

Looking forward to seeing how this plays out. 

 

Anyone See This Coming? The "Shame" of Coal Power for Data Centers

Many of us in the data center industry - or in the power business - have watched with some bemusement as utility-scale data center operators champion the energy efficient designs of their facilities on one hand, while seeking the lowest possible power costs on the other.  

Those low costs in rare cases come from sweet deals with municipal utilities with clean power generation portfolios - like hydro power in eastern Washington. But more often than not, cheap electricity comes with the environmental "price" of being sourced from coal-fired power plants.

Well, I guess Greenpeace is less amused, and certainly more forthright in criticizing this practice. According to the New York Times Green blog, the environmental group is using social networking to criticize Facebook for the location of their new data center in Oregon, and more particularly where local utility PacifiCorp sources the power that will serve the site.

Facebook defends the location as ideal for taking advantage of free cooling, and while I can't divulge everything I know about their operations, I can say that Facebook is in the vanguard of companies driving innovation in energy efficient IT equipment.

But Greenpeace has a point, and one that can be fairly extended to the utility-scale data center market as a whole. 

The resolution of this conundrum, in my view, is an increased partnership between the industry and utilities, addressing energy efficiency, demand response, and yes, power sourcing. The IT industry as a whole needs to engage proactively in these matters to avoid potentially damaging criticism now and in the future.

CB Richard Ellis - On Top Of Utility Issues for Data Centers

At the GSMI Green Data Center conference in San Diego yesterday, Jason Shepard, a Senior VP at CB Richard Ellis gave a great presentation on determining optimum sites for data centers.  He specializes in the Southern California region, but is part of a broader Technology Practice Group at CBRE.

Jason covered a lot of ground: industry trends (including the rise of collocation), the site selection process (avoiding natural hazard risk, communication service availability), and the evaluation of power reliability, availability, and costs.

He also put the power question in the context of "green considerations", raising all of the right questions about the power generation portfolio of a given utility. Married to water availability and the applicability of free cooling in a given climate zone, he made a compelling case for a careful evaluation of power based not only on costs today, but what it might look like in a carbon-regulated environment in the future.

One of his slides compared the power generation portfolios of five states, noting the relevant preponderance of fossil fuel-fired plants and the differing amounts of renewable and non-carbon sources. I was very pleased to hear him note that even within states there can be quite a difference in the generation portfolios of utilities.

In my white paper on the Carbon Content of Utility Power (available to consulting clients), I noted the wide disparity in power generation portfolios for utilities in California, with some having no coal-fired generation (the three big investor-owned utilities), and one getting almost half of its power from imports from coal plants (a municipal utility in Southern California).

If you believe that carbon regulations will raise prices in some fashion (and they will in my estimation, it's really just a matter of how much and how soon), then sourcing your power from a utility with a high-carbon power generation portfolio represents a cost risk.

Uptime Symposium Presentation

I enjoyed attending the Uptime Symposium last week in New York City, and will post some thoughts and reactions throughout this week.

In the meantime, I had lots of requests for copies of my presentation deck "The Carbon Content of Power: A Primer for Data Center Managers":

Download Uptime 5 17 10

While the content did draw some good discussion, and I did get positive feedback, I must apologize for delivery that was not up to my standards. 

In any event, the key messages are that the information sources for carbon content of utility-delivered power are very rudimentary - you should be sure to drill down to the level of your specific utility provider to get a good conversion factor.

Further, because data centers have essentially a unity load factor, managers are likely to overstate the carbon footprint of their facilities because the carbon content of electricity can be heavily time dependent.

If you are a consulting client of mine and would like to learn more, contact me for a copy of my latest white paper on this subject.

Cloud Computing: A Three-Tier Primer

Are you at all a novice when it comes to the subject of cloud computing?

Never fear, the New York Times is here, with a nice, taut feature today on the subject, written by Brad Stone and Ashlee Vance out of the San Francisco bureau.

They note that established companies are moving fairly slowly to cloud offerings from the likes of Amazon, Microsoft, and Google, but that startups (so prevalent in Silicon Valley) take to the model as essentially a standard practice.

Over at gigaom, Joe Weinman delves further into the potential market transformation of providing information technology as a "utility" service, riffing off of Nick Carr's seminal book from last year, "The Big Switch".  (I recommend the book, and his blog.)

The consensus of the two articles is that the move to the cloud won't be a "big bang" event, but more of an orderly transition led by small, nimble companies first, followed to some extent by established companies and organizations.

I personally wonder what impact this shift will have on utilities and the broader issue of energy use by IT. Clearly, the cloud players are building what I call "utility scale" data centers, with leading edge energy efficiency. But they're also siting them where utility costs are low, which often means the power is generated from coal (like in the Carolinas).

That speaks to a potentially positive contribution to the environment if everybody plays their cards right (and the cloud providers tend to the power supply side of the equation), but I wonder if we're simply deluding ourselves if we think we can lower the slope of the IT energy use curve.

My prediction is that the focus on IT energy use as a negative impact on the environment is going to intensify (even Greenpeace is on the case now). The IT industry will have to double their efforts to avoid this criticism, and utilities are in the enviable position of standing next to the people getting shot at. 

The best shield for utilities? Aggressively pursuing energy efficiency programs for data centers and IT.
 

Carbon Content of Power

Thought I'd make a small point that frames the debate about the carbon content of power. (I'm polishing a white paper on the subject, and will be delivering a presentation on same at the Uptime Forum in May in New York City.)

All of us, with very rare exceptions, get our power from the so-called "grid". Never mind that it might be better characterized as a radial network - there are no two-way streets and all roads come together at critical nodes!

If I happen to live next to a coal-fired power plant, then it can be reasonably surmised that 100% of the electrons coming into my home are from that station.

But here's the key point: if my utility owns that plant as well as a portfolio of others (perhaps including hydroelectric, natural gas, nuclear, etc.) my power use is more properly characterized as coming from that portfolio, with all of it's attendant attributes.

This is obvious in terms of the price you pay - just because you live next to a hydroelectric plant with low generation costs doesn't mean you pay less than the price of the generation supply portfolio.

Similarly, you should count the carbon content of the power delivered to your facility based on the generation portfolio of your utility, not on your location on the grid.

White Paper: Power Carbon Content

Working hard on a white paper that describes how best to go about evaluating the carbon content of the power delivered to you by your utility.

It's not as easy as it sounds, though there are a couple of places you can go to get a reasonable picture.  I'll point out the limitations of California's mandated Power Content Label, why the federal government information resources are only good as far as they go, and why data centers in particular may overreport their carbon emissions if they use the average carbon content figures from their utility.

And if this wasn't difficult enough, I'm going to take a stab at a methodology for accurately predicting power rates - again, the average rate figures provided by utilities are poor predictors of costs for high load factor customers like data centers. No promises on when this one will be done, as I expect the case study comparing California utility rates will require a pretty detailed spread sheet analysis.