Groves Green IT

SVLG's 2nd Data Center Energy Efficiency Summit

The 2^nd Data Center Energy Summit taking place on October 15, 2009 at Netapp corporate offices in Silicon Valley was really a feat.  Silicon Valley was in such a state of chaos in the early months of this year as the recession hit small and medium size businesses, that it was difficult to know whether there would be sufficiently interesting and completed pilots.  Yet, the Valley rallied and the event was full (400 plus attendees), the speaker podium compelling, and the audience was in good spirits. 

 

Speaking with Zen Kishimoto, PhD Principal Analyst, Green IT at AltaTera Research Network  who edited the case studies, he brought his own insights as he completed the task  he faced this year of compiling the results of the case studies (Teresa Tung of Accenture had this role last year):  “Like last year, we had a lot of participation and many excellent results.  We have fewer completed case studies than last year but will have several additional ones coming soon. One of the studies that attracted my attention was done by Power Assure. It proved the reduction of power consumption by IT gear can be achieved via hardware refresh and dynamic power management… Another pilot is the one conducted at an Intel data center, which measured temperatures at each server and reflected them to the overall floor air conditioning via wireless sensors. This seems to be the first attempt to link the condition of IT gear to the building management system Overall the reported technologies and practices are of high quality and I am sure the audience will find the results highly applicable to their own data centers.”

 

KC Mares of Megawatt Consulting and co-chair of this event,  said: “Worldwide, we spend $23 B as an industry on data center industry annually. That is the equivalent of five major nuclear power plants. Last year’s pilots resulted in an average of 56% ROI for end users. Our goal is to chop away at those huge numbers with our case studies.”

 

After a keynote speech from NetApp’s CEO, the audience heard an interview between Jeff Byron, Commissioner on the California Energy Commission and KC Mares.  Jeff said that California rate payers will spend $38 billion over next 10 years to reduce GHG but we can’t afford not to spend it.  He said the electricity sector contributes 23% of Ca’s GHG and all industrial sectors will be asked to reduce their 1990 levels by 2020.

 

This year, I played a background role in the Summit, participating in the leadership team that has met since January 2009 to bring the case studies to completion, created the public relations strategy for the SVLG Data Center Energy Efficiency Summit. These are some of the pilots I found most interesting:

 

1.  In the new Research Triangle Park NetApp data center, Joe Miller, site operations manager there,  said his infrastructure supports 369 IT racks, 1800 engineering racks and 30 engineering network racks. The 360 IT racks run at 6 kw per rack, whereas the 1800 engineering racks run at 12 kw/rack.  Custom air handlers for the NetApp data enter enabled them to develop a patented process to enable a cold room and hot aisle . Equipment draws the air where it needs to go, so  they don’t need to confine it too tightly. Hot aisle is open return in their pressure-controlled cooling architecture with two sub-systems: one of modulated temperature, and the other is to manage the dampers to modulate the air flow with the controlled fan speed of air handling units dedicated to the  cold room. Netapp  calls this “cold aisle containment with pressure controlled cooling”. The chiller plant operates at 55 degrees. The air side economizer means the chiller runs fewer hours than conventionally.

 

Typically if you elevate the chiller temperature, it increases fan power. But as long as it is below 80 degrees going into the front of the equipment, you can allow elevated chiller temperatures.  (The most efficient chiller is one that is off, Joe joked) And Netapp tries to keep it down. Cold aisles are contained and hot aisle are open so dampers can fill it out. Cost savings features described above resulted in a $50K per rack build expense rather than $150K per rack, the industry average. 

 

In the event of a chiller failure, there is the opportunity to ventilate. The number of hours that they cannot ventilate is a small risk, he assured the audience

 

2. Mukesh Khattar gave an interesting presentation on the case study he did at Oracle.  He started with the assumption: Variable Frequency Drives (VFDs) are required for hot air containment with any type of enclosure. But do we need VFDs if we don’t have enclosures, he asked of his team.  Enclosures do avoid hot/cold spot issues but the cost savings really comes from the FVDs.  He said that in a 5K sq foot data center, they managed a retrofit. The estimated cost of the project was $130K. They estimated energy savings of $27k per year, and a utility incentive of $33k.  The fan speed went down 50% with dynamic load matching. Oracle concluded that they could avoid the cost of curtains which brought the cost of the project down by $50K roughly. It made the ROI very attractive. One of the unanticipated outcomes is that you could not reduce fan speed below 50% - the fans turn off.

 

3. Joyce Dickerson reported on a very complex project undertaken at Stanford University campus. They discovered 70 different server rooms with a wide variety of purpose and use. They selected five to meter with Synapsense products, and looked among their 70 installations for one of each of these: cooling types using DX, CRAH units, Rittal water cooled, APC Inrow, and fan coil & building air conditioning system hybrids.  Across the campus they found that 1-3% of the floor space but 9 – 30% of the energy in each building were taken up by these server “closets”.  Their first conclusion was to always isolate the satellite servers from the building air conditioning systems: that saved the most money because air conditioners generally had to run an entire building or floor while just a small area required the cool air.

 

4. Pixar did a cold aisle containment case study and with simple blanking panels, panels and curtains, and air-dense foam, and had a payack of 3.7 months on costs to obtain an annual energy savings of $30K.  Be sure to read that case study.

 

5. Ralph Renne of NetApp hosted a case study whose objective was to move their focus from temperature control to pressure control. To ensure that cold aisles will not be oversupplied or undersupplied,  Renne concluded that temperature sensor placement becomes crucial.   Temperature control does not guarantee that the volume of air is being delivered but pressure control does, and that pressure control also delivered more air into their cold aisle. 

 

Andrew Fanara of the EPA, who attended last year was here again this year because in his own words, “the activity around energy efficiency in data centers is more specific this year, moving away from rhetoric to action.  Companies making investments and gaining knowledge are better positioned as the economy turns around. SVLG has to be admired for pulling this event together in a time when most companies were pulling inward and cutting out any “extra projects”, yet as I’ve seen today, there is tremendous activity still going on.”

Paying tribute to MJB and acknowledging the power of peer to peer networks

First things first:  Mark Bramfitt of PG&E’s data center energy efficiency team, is leaving the company to pursue consulting opportunities.  he announced this last Thursday at the SVLG Data Center Energy Efficiency Summit 2009 at NetApp.  Mark has been my source of knowledge on energy efficiency, mechanical engineering and utility-programs related topics. He has been generous in educating me.  He has also been generous (with all of us) in his role as spokesperson for energy efficiency initiatives that utility companies can choose to offer.  His leadership across the industry undisputed.  There is no one to step into his shoes and he leaves a void.  Thanks so much, Mark!

I have more to say about the Summit but I'll include that in a future blog - just that it was a great event and my notes are on a different computer.  I wanted to take this opportunity to say that I had the honor of addressing the Site Uptime Network on Tuesday, October 20^th, in Atlanta Georgia.  I was there to present to the audience on research I am conducting, on the subject of “Awareness and attitudes toward water conservation in data centers.” I gave an overview of drought conditions in the USA and around the world, and showed a map that indicated that the predicted drought areas in north America overlap to a large extent with the areas where drought might occur. 

I became very interested in this topic when I realized at the Govenergy conference in August that energy efficiency initiatives across the federal sector in the USA included water conservation milestones.  (see earlier blog).  Although data centers were not expressly mentioned, 15% reduction in water usage from 2007 levels by 2015 in all GSA buildings indicated to me that the private sector data center operators should be implementing water strategies to coordinate with their energy strategies – and I didn’t know whether it was being done.

So I developed a 39 question survey together with Dr. Bob Sullivan and David Rosenberg and I am doing the research now. Results will be published in early 2010. Those who fill out the survey completely will be given the report at no charge.  The report will be for sale to other interested parties.  Contact me if you have questions about participating or sponsoring the research.  Your data center must draw at least 5MW of power to be included. Go to "email me" in the right hand column to reach me.

You can see some of the  North American Site Uptime Network here.  Although I had known about the group for a few years, I didn't know the following until a few days ago:

Network members average a 50% reduction in site failures during their first five years of membership.AND Network members average a 50% decrease in incident severity during their first five years of membership. These numbers must be quite comforting and well worth the $15K annual membership fees.

Ken Brill presented a slide that he titled Site Resilience = Tier plus sustainability.  He said Tiers certification is about concurrent maintenance and fault tolerance in contrast to operational sustainability which is about frequency of unexpected failures and duration of failure and ability to recover.   He also went over five tier topologies that are needed to achieve operational resilience. It was a masterful roadmap that was intuitive and helpful. 

Members are divided into North America and EMEA.  The North America group has about 100 members but some of them have asked to remain anonymous so a complete list is not readily available outside the group.  It has been going for 17 years. One of the earliest members, American Express, has been participating since 1993.  The members are divided into groups that generally don’t lump competitors in the marketplace together. So with five different groups going, there can be five corporate memberships in insurance, auto, banking, etc., without a breach of confidentiality. New members are voted into the group. 

 

Participation includes not just fees but active contribution. These include opening your data center to site visits of other members, and notification of abnormal incident reports, a minimum of four per year.I had the opportunity to listen in to a debriefing after a visit to a member facility and it was very illuminating and refreshing to see the peer to peer sharing that took place.

It seems to me that $15K is a small price to pay to share in such large risk mitigation.  Have a Green Day!

 

.

Nuclear power to fuel data centers?????

Last night I went to hear Stewart Brand (one of the original authors of The Whole Earth Catalog) discuss the highlights of his latest book at the LongNow Foundation lecture in San Francisco. His new book is titled Whole Earth Discipline.  You can order it here if you like after reading about him and his fascinating conclusions. 

Stewart has long been one of the environmental literati, who hobnobs with Paul Hawkins, Amory Lovins, and others who started the movement toward appreciating and protecting the earth. And as he admits in the first few minutes of his talk, it sounds like heresy that he is now an advocate of nuclear power.  But, he says, the evidence just kept growing over the last four years as he did research on global warming, that the scope of the warming is so acute and urgent, that no small panacea would suffice as solution.

Realizing that he would meet skeptics that would want to see the data that persuaded his new thinking, Stewart decided to put all his research notes that he deemed of value (he says he discarded the dead ends) online for all of us to peruse here.  

Stewart's book is 1/3 about warming as a result of fossil fuels, 1/3 about geo-engineering - or recombining our land use and land use laws, so that more of it can remain "wild" natural, or pristine (I am not sure which adjective he used) but un-cultivated, and 1/3 about the growth of cities in the southern half of the globe. For the purposes of this blog, I am concentrating on the first 1/3.

"Brand declared that geological sequestering of nuclear waste has been proven practical and safe by the ten years of experience at the WIPP in New Mexico, and he paraded a series of new "microreactor" designs that offer a clean path for distributed micropower, especially in developing countries." it says on the LongNow website, and it was the photos of various very small nuclear power plants using, for example, thorium instead of uranium, that were fascinating to me. I'd never heard of thorium before yesterday. But according to Brand, a small thorium power plant can be dug underneath an industrial building and can remain self contained without ever removing any waste/residue, sidestepping the issue of where to put all the nuclear waste that has troubled us for the last half century.  Here is another opinion on the benefits of thorium reactors. 

It seems redundant to say I don't know anything about this topic, only that if Stewart Brand wants to stake his brand on persuading us that we should re-think nuclear power in the context of green, I am willing to give it additional thought. I hope you will too.  When Kevin Kelly (also a LongNow founder), Stewart's longtime colleague who was interviewing him after the formal speech, asked Stewart how he was able to maintain dear and valued friendships when his thoughts on controversial issues change 180 degrees, Stewart replied that he tries to learn from every conversation instead of sticking to his beliefs.  Another simple truth from a seer worth listening to.  Have a green day!

Understanding power distribution with IEM

1. Ed, at last week’s CFRT meeting you gave a brief description of your company’s history and allowed us to view your manufacturing process in tours. Could you describe your company mission and a little history?  .

IEM  is an independent electrical equipment builder with history back to 1950 when Frank Howe started producing equipment in a barn in Fremont.  He remained fiercely independent during industry consolidations in the later 1950s and developed an approach to the business that allowed him to remain a market share leader in Northern California until the 1980s.  His innovative process and very early adoption of CAD/CAM methods were key to establishing that leadership in Silicon Valley.

We have tried to continue that independent spirit and to implement new technology to grow our business beyond the local market with a focus on customers who want and need specialized solutions to their power distribution needs.  The efforts of our 135 employees have resulted in revenues of over $50M last year, and we are poised for significant growth.

2.  As you addressed our group, you spoke about the energy efficiencies of using medium voltage power? Could you recap that here? 

Most data centers require large amounts of power that the utility company supplies at medium voltage which is technically defined as 2,400 to 38,000 volts.  Data centers are normally provided with power between 12,000 to 34,500 volts.  The normal distribution strategy is to transform the utility power to a lower voltage (generally 480 volts) and to distribute large blocks of power at that lower voltage.  

Our view of power efficiency in data centers is focused on what we know — power distribution.  Every time electrical power is transformed in terms of voltage or from AC to DC or DC to AC, there are losses which generally also have a heat impact where they occur.  Our strategy is simple:  minimize the number of required voltage or system changes and the result will be fewer losses.  By using medium voltage further into the system (even onto the raised floor) that goal can be accomplished and gains in efficiency of 8 to 15% can be realized by the data center operator.

3. Most data center energy efficiency think of Direct Current (DC) when they think of power greater than 220 Volts. Could you describe why it is not necessary in your vision of energy efficiency?

DC systems over 48V are not widely used in the sites that we have seen, so I am not familiar with all of the implications of higher DC voltage systems.  DC power does have advantages in regard to synchronizing various sources more easily than AC; and larger AC to DC conversion devices (an AC to DC conversion is ultimately required within the server with an AC system) may well prove more efficient.  At the same time, DC power has its own set of challenges since it is far more difficult to interrupt DC power than AC and simple things like the plugs to connect servers to the system need to take this fact into account.  In any case, the two strategies are not necessarily in conflict since the final conversion can be made from the medium voltage AC system to either a AC or DC floor distribution system.

4. How long do you think it will require to bring the 4180 power into a mainstream data center as a pilot, and will it take a decade or longer to gain acceptance as an industry standard?

There is an operating 4160V system in a major data center which is in full operation today.  The major issue has been the availability of distribution equipment that is more suitable to the data center environment and the associated operating needs.  We expect that there will be significant interest now that such equipment is available, and I doubt whether it will take more than a few years for MV to be a common distribution voltage.  The challenge for full implementation will likely be the availability of MV UPS systems that don’t require transformers.

5. I understand that Square D, Siemens, GE and Cutler Hammer (Eaton) are other significant players in your current existing product lines. When you roll out your higher voltage cabinets, will that competitive landscape be significantly different? Who else sells solutions along those lines?

All major manufacturers have existing MV products.  All of those manufacturers have a strong market presence and loyal users.  The manufacturers’ challenge will likely be to adapt their existing MV products which were designed for utilities and heavy industrial users to meet the new needs of the data center markets.  Most of those manufacturers also have UPS companies which have a significant investment in 480V systems which will require major changes to adapt to a MV distribution system.  I am not certain how those manufacturers will respond:  whether they will respond with new products or try to resist any changes.  We also know that at least some of the data centers will continue with their current 480V or 600V systems for expansion and new sites rather than re-train their personnel.  Other users — and we hope there will be many of them — will be eager to get the operational advantages of a lower cost system that uses MV distribution. 

HAVE A GREEN DAY!

 

3 new opportunities for data center EE training & certification

DatacenterDynamics is set to offer data center design certification in San Francisco. Their first course in San Francisco took place earlier this summer, and it seems there is enough demand to offer the course a few times annually going forward.  You can find out more about the courses here

Meanwhile in Kevin Heslin's blog, I read that The Uptime Institute is offering training of a similar vein. This is what I read on their website:

"This course responds to a) the fact that the Tier Classification System is increasingly propagated in data center projects across the globe and b) consistent feedback from the engineering community that the Tier Classification white paper in and of itself does not provide a sufficient initiation point for a detailed design solution.

This will be an intensive 3-day course focusing on materials and concepts of a technical nature, which will culminate in an examination. The cost of the course is $4,985. For individuals that pass the examination, the Uptime Institute will award an Accredited Tier Designer endorsement. The course will be administered by Uptime Institute Faculty with field experience in the practical application of the Tiers. Furthermore, each Faculty will be a co-author of the Tier Classification and Operational Sustainability white papers and a Tier Certification Authority." 

In a third effort to recognize the difficult task of knowing everything about infrastructure and IT simultaneously, the Department of Energy is working on a DOE Certified Energy Practitioner (CEP) Program for Data Centers as an adjunct to their DC-Pro tool for assessing energy efficiency in the data center. Magnus Herrlin describes the curriculum in great detail here.

This is all good news, isn't it?

By the way, I will be speaking at this Site Uptime Network event, October 18 - 20 in Atlanta, GA. My topic will be "Water: the next oil?" and include an overview of the drought conditions that are raising awareness about availability and pricingof water. Are you interested in the results of my study? It will be published in Q1 2010.  Have a green day!

GovEnergy Conference last week: A huge success!

Before I start to describe all that I learned at the above conference, have a look at this Department of Energy page  under 'Resources'.  You will see a presentation created for CEO/CIO audiences, to engage them in data center energy efficiency.  I created this with colleagues at NREL  the National Renewable Energy Laboratory. You can view it as a PDF or as a PPT.

Govenergy had over 3000 participants this year. The event, August 9 - 12, 2009 in beautiful downtown Providence, Rhode Island, promised and delivered on the following:

• Gain insight on reducing Federal agency energy usage and cost, as mandated by the Energy Policy Act of 2005, Executive Order 13423, EISA 2007 and additional Federal guidance.
• Access the tools, techniques and best practices needed for meeting both day-to-day and long-term energy management goals.
• Develop effective partnerships amongst Federal energy professionals.

Here is the significance of the first bullet above:
Executive Order 13423 of 2007 and EISA 2007 set new federal energy goals:

• Cut energy use (compared to 2003) by 3% per year in 2007 through 2015
• Increase use of renewable energy to not less than 3% of total electricity use in 2007 – 09, not less than 5% in 2010 – 12, and not less than 7.5% in 2013 and thereafter, with at least half from new sources in each year
• Reduce water use by 2% per year, 2008 – 2015

One of the most telling comments came from another attendee during a lunch break and I cannot even remember who said it.  He said that the federal government had tried a "bottoms up" approach for the last several years toward energy efficiency, and had few successes, and was now trying a "top down" approach with mandates. 

Coming from the West Coast, I'd be the first to admit that I have very little clue what goes on in Washington D.C. But I have formed impressions which may have been accurate or not.  Let's just say I was very impressed at the vision, the magnitude of effort, and dedication in the program managers that I met at this event, all working fairly cohesively toward an Energy Efficiency goal.

The key player, in many cases, is the Federal Energy Management Program (FEMP) within the Department of Energy. FEMP has selected 16 companies to participate in their Energy Savings Performance Contracts. Those 16 companies are called 'ESCO's

"The ESCO guarantees that the energy improvements will result in a specified level of annual cost savings to the Federal agency customer. The cost savings are also guaranteed to be sufficient to allow the agency to pay the ESCO for its work over the term of the contract. After the contract ends, all additional cost savings accrue to the agency. Contract terms up to 25 years are allowed."

It is estimated that auditing all their work will require 25,000 audits annually in government buildings.  It was predicted that an energy auditor could complete just 75 audits per year with all the documentation required so that is a huge investment in cutting energy use!

Ok, so what does this mean for you, the reader? it means that if you want to retrofit a data center housed in a federal building, there is a process to finance, through FEMP's assistance, capital costs over a 25 year period.  you can do a lot of work under those conditions, don't you agree?  Doesn't it make you feel good to know that the federal data centers are going to be funded to be energy efficient? I am very encouraged! 

Have a green day!

Just thinking about trickle-up innovation

I've been an avid Wired magazine fan since it was first published. Some of the most lucid writing has appeared on their pages, even articles that are unrelated to technology. One article comes to mind immediately about a teacher in Arizona putting four "at risk" youth through higher level math courses that resulted in offers for college scholarships for then. Thank you Kevin Kelly for pursuing journalism at a very high level there. Another article about Luis Ahn at MITgot me thinking about the web populace as pattern recognition experts in progress who, as an alternative to suduko and jigsaw puzzles, focus on facial patterns so that those missing children on milk boxes can be identified and returned to their parents. 

As often as not the articles challenge a way of thinking as much as a description of the new and noteworthy.  In March of this year, I read about The Netbook Effect by Clive Thompson. The subtext reads: "Dinky keyboard. Slow chip. Tiny hard drive. And users are going crazy for them. How cheap little laptops hit the big time."

Although it is counter intuitive, Ithink that there is a potential analogy to data centers in the above scenario: i.e.,  we want less not more. Can that possibly be?

The inventor, Mary Lou Jepsen, is quoted as saying, "We started inventing technology for the bottom of the pyramid but the top of the pyramid wants it too." Clive Thompson goes on to write,  "This bit of trickle up innovation, this netbook, might well reshape thecomputer industry- if it doesn't kill it first...Netbooks have ended the performance wars."

(An aside: I don't think it will kill it. In fact, more cheap computers everywhere expands the market exponentially - again. After just reading David Hitz'  book "How to Castrate a Bull" (very well written, funny, incisive and honest!) an enterprising reader concludes, "take the risk and see yourself at the beginning of an even larger wave to ride, not fighting over a smaller and smaller pond."

Other interesting tidbits Thompson provided: "The cloud tail is wagging the hardware dog" and "It's the rise of the very small machines."

Fast forward to my own interpretation of the consequences of this type of thinking: what if the concept of less not more, smaller not bigger, and new larger ASHRAE envelope for data center cooling and humidity means build 'em small, cheap, and ubiquitous. Maybe the neighborhood data center is the size of a composter in the far end corner of my backyard and fits around the utility pole.

Without going back to class I can't see how to put this into coherent engineer-speak. However, looking at the migration of freezing raised floor space to server closet and half-way back again leads me to believe that a future design can be smaller, cheaper, and even more ubiquitous.

We don't have beautiful data centers on the West Coast of the USA. I've seen photos of some beautiful ones built for New England landscapes, but out here they are lucky to be as pretty as an airplane hangar.  Why? I ask. "Security". "Cost". Always the answers I get. Instead of thinking in terms of a huge infrastructure to support miniaturization at the edge, think in terms of distributed power and answer:

1. Can we design data center ubiquity to lower the security risk?

2. Can we use the principles of the netbook to lower the cost of data centers?  

Have a green day.

Andy Lawrence of 451 speaks about Carbon Trading & Data Centers

Andy: I understand you are doing research on the impact of carbon trading on data centers.

1. What will be the focus of your study? US and Europe? Elsewhere as well?

 Over the past two years, there has been a lot of talk about regulation aiming at reducing datacenter energy use. In the US this partly stems from the findings of the 2007 EPA report to Congress, which highlighted the growing consumption of energy by datacenters, and partly because both the Dept of Energy and the EPA are starting to gather a lot of serious data – the kind of data you would need for a baseline for any regulation. In Europe, the situation is more clear cut. The European officials behind the European Code of Conduct have stated bluntly that if energy use by datacenters is not controlled, it will be regulated.

 There are, of course, many types of regulation. We’re focusing on Carbon Cap and Trade in this report because that is already happening in some areas, and because this type of scheme is favored by most economists and politicians. Our primary focus is on the USA and Europe, especially the UK, although we are keeping an eye on developments elsewhere. We are focusing on these geographies for two reasons. First, this is where most of our clients and their customers are (The 451 Group serves the IT industry and its leading edge clients; our sister company Tier1 Research servers the commercial datacenter industry); and second, it is where most datacenters are, where there is most concern over legislation, and where most significant legislation is likely to come from, at least initially.

2. What existing and pending legislation are driving demand for this report?

 This an interesting question, since the situation is somewhat confused, and it’s certainly very fluid. To a point, some of the demand for information among CIOs is based on some misunderstandings about Cap and Trade and even some scaremongering. Some IT companies have been stoking up concerns among CIOs, hoping to open a dialogue that will lead to some sales.

 As it stands, none of the big cap and trade schemes in Europe (the Emissions Trading System, or ETS) or in North America (the Western Climate Initiative Cap-and-Trade Program, the Regional Greenhouse Gas Initiative, and the Midwestern Greenhouse Gas Reduction Accord and the Canadian Carbon Cap and Trade scheme) really affect datacenters directly, since they are targeted at the biggest carbon emitters – power companies, cement companies and the like.  The American Clean Energy and Security Act , which has just been passed by the House, will probably not directly affect datacenters either, although this could change.

But there is a big rider on this: The regulators mostly won’t have to pass new laws to have industries redefined as “Energy Intensive”. Datacenters are already being seen as that by many officials. That can be a double whammy, because energy intensive industries area also legislated by the Clean Air Act.

And there is also a strong likelihood of more regulation. In the UK, for example, the Carbon Reduction Commitment (CRC) mandates all businesses that use more than 6,000 Megawatt hours a year of electricity to enter a cap and trade scheme from 2010. If you have a datacenter that draws more 684 KW, you are already over the limit.

 3. How urgent is this? Is this a 6 months or 18 months strategic imperative?

 The urgency of this depends on what you do. If you supply systems, software, datacenter equipment, or services, I think you need to be thinking about any legislation might affect your products now. For example, carbon cap and trade requires carbon measurement, and maybe carbon accounting. Carbon measurement requires energy measurement. And requires a lot of thinking about how you will measure, where you will measure, what you report, and what you want to do about it.

 Datacenter operators need to be thinking about all this, and they need to understand what their customers are likely to ask for. Suppose a customer asks for the ability to account for the carbon and energy use associated with a service? Suppose the CFO asks for a carbon  and energy projection for the next five years. This stuff isn’t easy. The simple answer is, Don’t expect anything dramatic in the short term, but expect carbon management to eventually become routine.

 Mike Manos, former head of Microsoft’s datacenters and now at Digital Realty, recently put it like this: “I would bet that you as a Data Center Manager have not been thinking of this, that your CIO has not thought about this, the head of your facilities group has not thought about this.  First you need to start driving awareness to this issue.    Next we should heed to a call to arms.”

4. Of course I understand this has significance for large data centers. What is the smallest data center that this will impact?

  It's difficult to say, since no proposed cap and trade legislation is targeted at the datacenter per se, but at overall energy/carbon use.  A first step is get an understanding of overall corporate consumption by function, by geography, by department. That will show how much energy/carbon the datacenter is responsible for.  It can range from 10% to 75%. In terms of size, the UK law sets a cut off point around 800 KW for the whole business.

 A second point to consider is that even if Cap and Trade doesn’t require direct registration, it will affect energy prices. Carbon tariffs, dynamic pricing and mandated renewable energy investments, and oil and gas shortages, will all combine to increase electricity prices, and to increase volatility. Big datacenters may need to hedge. The short answer is that businesses need to get on top on this. It’s going to be a business risk.

 5. So far, in your research, what seems to have jolted you a bit? Surprises? Ho-hums?

 First, we have been struck by the willingness of many IT vendors and consultants to go to market with advice, services and products around this area, without really understanding it. But I expect this will change in the two years ahead, as it becomes more strategic. The top tier IT companies have on the whole taken this very seriously and are developing a lot of real expertise.

 Second, this is a very difficult area for end user organizations to crack. There are really three different disciplines and vocabularies that a company needs to understand: First, there’s IT – for a lot of us, that’s the easy bit; Facilities, which includes power and cooling; and third, carbon accounting and markets. If you also understand how politics and legislation works, that helps a bit too.

 Another point to make is that Carbon Cap and Trade systems are inherently difficult to get right, and to make work – witness the machinations of the ETS and some of the lobbying in the UK ahead of the CRC. They are a kind of economic experiment, with so many variables that the results can be unpredictable. So expect any schemes to be controversial, and, to some, unfair, and expect a lot of “gaming”. The winners may not necessarily be those that are the greenest, but those who have learned to play the system.

6. Reactions you are getting from data center operations staff – willingness to look for creative solutions or just keep on paying utility bills?

 As an industry, datacenter operators take quite a hard headed approach. They have come round quite fast to the importance of energy conservation, because its really about saving money.  But carbon is another matter – in Europe it plays fairly well, but in the US it is low on the agenda. On the whole, there is some evidence that businesses would rather be regulated as a group, so that the more conscientious are not a commercial disadvantage. But it’s unclear.  The most forward looking commercial datacenter operators are the least concerned. They see this as an opportunity exploit their efficiency. Theoretically, they can lower their costs if their own energy costs are lower.

 7.  Is there really a way to get away from carbon trading? Offshoring to Asia for example? What do you see?

 The extent to which carbon trading or accounting can be avoided will depend on the framing of the laws and on the jurisdiction. And the extent to which anyone will actually want to avoid the carbon trading will depend on the price of the carbon and availability of permits etc. To date, the financial burden of being in a carbon trading scheme has been so light that it hasn’t been worth going to great lengths to evade responsibility. But yes, until or unless some strict laws come in that track emissions through the supply chain, it will always be possible to outsource some responsibility.

 Under the UK's CRC, it is certainly possible to avoid responsibility for carbon emissions, because it’s based on direct emissions – supply chain partners don’t count.  So outsourcing datacenter services, even within the country, will take it off the books.

 This has led to some grumbling about the CRC, and suggestions that it will lead to some offshoring or “carbon laundering”…the use of “off balance sheet” businesses to take on the carbon load. But this isn’t necessarily a bad thing, if they are run in a carbon efficient way. This is, in fact, an opportunity for efficient datacenter operators.

Thanks Andy! P.S. At Datacenter Dynamics in San Francisco, just last Friday, Zahl Limbuwala of the British Computer Society gave a presentation that reached similar conclusions on the topic of offshoring.  But Zahl really thought the CRC legislation was very off target - punishing and unfair to those companies that had already started their EE initiatives and were now forced to work on more difficult tasks than those companies just starting out with 2008 as their baseline.  He thought Americans should be careful not to be caught "asleep at the wheel" while decisions are made about data centers in legislative bodies without careful analysis in advance.  Try then, to have a Green DAY!

 

 

Energy Summit and the Institute for Energy Efficiency at UC Santa Barbara

This blog includes notes from my interview with DanColbert, Executive Director of the IEE.    

1. Dan: you just held a very successful 1^st Summit on Energy Efficiency in Santa Barbara, CA in late May.  I was there for the entire two days and I found it to be valuable for its insights, its predictions, and its networking.  Do you feel you reached the milestones you had hoped? And what was the highlight for you?

Very much so.  The mission of this Summit was – in addition to creating greater awareness of the Institute for Energy Efficiency at UCSB and what we’re doing – to raise consciousness about the need for more and new technologies that enable greater energy efficiency.

2. When I hear most people talk about energy efficiency, one of the most compelling factors they cite – correctly so, as far as it goes – is that the technology is ready now; we don’t have to wait, as we do for many others, e.g., plug-in electric cars (battery technology), photovoltaics (lower cost), carbon capture and sequestration, etc.  Technologies for energy efficiency are here now, and we simply have to implement them.

This view is in a large sense correct, and has led on one hand to groups like the Energy Efficiency Center at UC-Davis, which does the important engineering work of studying systems of already-developed technologies.  Another outflow of this view – exemplified by work at the Precourt Institute at Stanford – is to study behavioral issues relating to the interaction between people and existing technologies.  Again, this is important work.

Both groups, however, focus primarily on existing technologies at a systems or behavioral level, while the Institute for Energy Efficiency is focused on pushing the boundaries to research and develop new technologies for energy efficiency.  This is critical because realizing the full scope of energy efficiency requires new technologies.  So, in order to make this point most effectively, we brought to the Summit speakers who have direct experience – whose job it is to know – what new technologies are needed in energy efficiency, what the state of the technology and markets are, what policies are needed or are getting in the way, to fully realize the promise of energy efficiency.  Without new, better, cheaper technologies, only a small fraction of this promise will be met, even with the best systems engineering and behavior studies in the world.  In this way, our Institute, Davis, and Precourt are highly complementary.

For me, the highlight of the Summit was the town-hall meeting, led by Forrest Sawyer.  It was a dynamic session, and the participation was great.  In particular, it was fantastic to get a great question from Mr. Naugles’ Grade 7 Honors Science Class, Fremont School, Oxnard, CA.
It’s so important to get kids excited and involved in this effort!

3. Readers of GrovesGreenIT are interested in data center energy efficiency in particular.  Could you explain what Fred Chong’s group is working on?

As computing and data storage increasingly become globally available, public utilities, the proliferation of large numbers of servers, and massive data centers will have a substantial energy footprint in our future.  The typical server consumes as much energy in one year as an SUV. Worldwide, businesses now spend $30 billion to power their data centers, and that cost is growing rapidly.  Energy expenditures have now become more significant than the cost of the equipment, making energy efficiency of critical importance to our future information technology infrastructure and natural environment. The Greenscale Center for Energy-Efficient Computing serves as a focal point for a vertically-integrated effort to drive key energy-efficient technologies in computing and electronics, and couples horizontally to the Buildings & Design group through its work on data centers.  Collectively, these technologies address very significant near-term and long-term energy challenges.

 Some projects within the Computing Solutions Group of the Institute, led by Prof. Fred Chong, are:

Ø       Emerging Technologies for Energy-Proportional Computation

Ø       Cooling Technologies

Ø       Energy-Aware Computation

Ø       Life-Cycle Analysis and Policy

Ø       Wireless Networking

Finally, cyber-security is an under-recognized area of energy efficiency.  Last year, $3 billion in electricity cost was wasted just by spam.  More importantly, as the smart grid is developed, it will be increasingly vulnerable to cyber-attack; we must develop solutions to enable and protect such energy-saving systems as the smart grid, else they aren’t truly available ways to save energy.  Fred’s group is a leader in this research area.

4. One of the conclusions that was hammered into the audience on Day 2, the Policy Discussion day, is that now is the best time ever to be involved in energy efficiency.  Do you have anecdotes from your own experience on the ease with which this topic is being discussed in Washington DC and elsewhere at a national level?

Like you, I have attended several conferences on EE.  One thing that has interested me is that to a great many – especially those on the utility side – energy efficiency is equivalent to “demand-response.”  Another interesting aspect is how slow – although it is now accelerating – businesses have been to understand the financial benefits of energy efficiency.  There appears to be a fascinating psychological barrier (psychologists should study this!) to adopting an energy efficiency technology by a General Manager of a business unit when he or she thinks in terms of payback time. Historically, they require less than 24 months – more often 18 months! – payback before considering adopting it.  Requiring 18 months payback time is equivalent to an internal rate of return of 67% (!), which would be well above any CFO’s investment threshold.  Furthermore, this is just accounting for savings in energy costs – it does not account for productivity gains, which study after study have shown typically far outweigh direct energy cost savings.  So, the message is to think in terms of return on investment, rather than payback time.  Simple, but important.

More importantly, though, what has interested me is the very broad acceptance of the importance of implementing energy-efficiency technologies based only on the assumption of currently-available technology!  This view misses most of the e.e. boat:  much more will be gained by improvements in technology, and through wholly new technologies that save energy.  Finally, there is little awareness of unintended consequences. 

At the Institute for Energy Efficiency, we are studying the so-called “rebound effect” (aka Jevon’s paradox), which shows that in certain cases – depending on technology, market, and policy conditions – use of more efficient technologies actually causes greater overall use of energy (e.g., more efficient steam engines at the start of the industrial revolution caused hugely more coal to be burned).  This is a complex situation, in which technology, policy and economics all need to be considered in concert to fully understand the issues and fulfill the potential. HAVE A GREEN DAY!

 

 

Speaking from California: Interview with California State CTO PK Agarwal

I recently had the chance to speak with PK Agarwal, CTO of California.  In this role PK manages the State of California Consolidated Data Center used by various California governmental departments.  California has two large general purpose data centers and many large departments have their own dedicated data centers.  The State of California has 130 operating departments, boards, and commissions.

In a typical month, there are six million unique visitors to California state government websites.  Due to the recession, the demand for state services has skyrocketed and more and more people are looking for employment related services and other forms of social and financial assistance. 

California is going through a consolidation of IT and over time we will see a lot fewer networks and data centers.  When you remember that California has about 37 million people, and a GDP equal to the 8^th largest country in the world, then you understand the magnitude of the information to be processed. 

 DG:  I really enjoyed the talk you gave about multi-generational work forces (earlier this year at the Teladata conference in Santa Clara), and their various approaches to IT and social networking tools at work.  Could you summarize your thoughts about the younger generations (under 40) for the readers of this blog?

PK: The younger generation, historically sheltered by a strong economy, is now probably feeling some pain due to the US unemployment situation. This may result in a new appreciation and greater receptivity to older workers’ advice based on years of experience. Here is a critical issue worth considering.  Currently, 75% of California State employee managers are baby boomers, between the ages of 50 – 65. California’s managers in this age group are predicted to drop to 50% in 2014, leaving the baby boomer’s culture as the dominant management style. 

 However, in 2014, the dominant demographic group of rank and file workers will be Generation Y. There is the likelihood of a culture clash between the managers being dominantly baby boomers and workers dominantly generation Y.

We have to be sensitive to these culture differences.  You can make young people work but you can’t force satisfaction and contentment on them.  One big cultural issue is that these young kids want to communicate frequently and want continuous feedback as a part of their work environment.  This does not make them less productive or less security conscious.  It is just a different way of communicating.  So, we need to adapt to their needs.  Tools change, habits change, culture changes, that’s life.  For example, Gov Schwarzenegger now has a Facebook profile, and a Twitter account.  Both of these actions send a message:  social networks have a legitimate and recognizable value.  Also, President Obama’s staff has recently negotiated terms with four social networking sites so that there is clarity regarding the ownership of data on social networking sites.

DG: Forgive me for being blunt, but when I go to this website, http://www.cio.ca.gov/ it still looks like a traditional website for traditional people.  How can this be attractive to the under 40s we are trying to engage in state government? 

PK:  The primary customers for our CIO website are state staff and the IT vendor community.  Both of these groups are dominantly baby boomers and have some very specific data and information needs.  Accordingly this website is designed with that need and demographic in mind.  On the other hand, if you look at www.ca.gov, it has a very different look and feel.  We recognize that California’s increasingly younger and diverse population wants a certain ‘website feel’. We are also working on a version of www.ca.gov that is intended for our youths and has been designed by local high school students.

 DG: When do you think that the California website www.ca.gov above will include facebook, twitter, 2nd Life or other kinds of new media as part of the content? (for employees and state residents)

 PK: We are already using many of the tools you mentioned.  Many of departments are incorporating web 2.0 tools as a part of their communication and service delivery strategy.  However it is important to note that government websites cater to a very wide array of demographics.  Accordingly our websites are designed with an understanding that we will be serving power users as well as digitally challenged.  It is somewhat easier for a company such as Amazon because they can choose a single channel to serve their customers.  With a government entity, we have to be multi-channel because one size will not fit for all for our constituents.  We will continue to have walk-in, mail, telephone, FAX, and Internet as the service options.  www.ca.gov is the entry point into the State system but you can go directly to a department as well.  How do we get you from www.ca.gov to where you want to be?  We try to get you there in the fewest number of clicks.  Plus, we have seasonal variation in our web strategy.  During March and April (when it is the tax season), a lot of people want to deal with tax issues.  Another popular service all year round is that of automobile registration.  Lately we have been seeing a lot of activity around employment, social services, or programs for refinancing, etc.   California's first lady, Maria Shriver, also sponsored a website called www.weconnect.net and that is also getting a lot of traction with people in need.

DG:  Budget constraints aside, how do you translate your knowledge into a better place to work and a better site to resource?

 PK:  We continue to use technology to better serve Californians.  Come back around mid-June and look at www.ca.gov.  You will see our new website design intended to improve the user experience.  We have been thrifty and careful, but really focusing on the usability and the customer experience.  By contrast, our 2006 design would look primitive. 

 Also, we want to ensure that transactions are processed on our websites efficiency and quickly.  Unlike a commercial website stickiness is not as critical to us.  Often people want to conduct their transaction such as registering their car quickly and that is all they care about on that particular day.  So we try to find a balance between “bells and whistles” and functionality. 

 DG: Last question, PK:  I am sure that all the readers of this blog care a lot about your operations focus:  on how you perceive Green IT and sustainable computing to be a significant part of your agenda.  Are you going to recommend a 30% reduction in energy use throughout State data centers as the federal government has done?

 PK:  Yes we are very “green” conscious in our computing practices.  As you know our governor is recognized globally as a pioneer in setting the agenda for climate change and energy issues.  For instance we reduced the utilization of natural gas in one of our data centers by 90% and electricity usage by 9%.  We are looking at consolidating many of our data centers which should give us a lot to brag about in the area of energy efficiency.  Server consolidation and virtualization is among our key strategies for energy conservation.

DG: Thanks so very much PK.  Have a GREEN Day!