As I sit on the 8^th floor of a hotel in downtown Vancouver between a vacation and meetings in Canada for some new data center projects, I look out of my hotel room at the lovely Vancouver sky and skyline and notice the many living roofs. About half of the dozens of rooftops I can see are living, and as I walk the downtown area, I see many more from other buildings and hills and even the sidewalk. Why is it that this city has so many living roofs? I notice that many of the living roofs were “unintentional” living roofs, where a little dust builds up on the gravel or membrane roof, creating soil for grasses to grow, the they do in this ideal growing climate. Others were intentional. So, my point today is, if so many roofs are left to nature to create a growing medium and grow grasses, without man’s design intent, and left in this natural living method, than why are we so afraid to build living roofs in other places? Most say it’s because they fear that a living roof will leak. Well, I can tell you, if all of these living roofs on Vancouver were leaking, you’d know within a day, as it has rained here every summer day for a week now.

Studies show that living roofs are actually less likely to leak. Why? It’s the same roof construction as a typical roof but only better. And if it leaks, you repair it the same as any other roof, just a little plant or soil matter to first slide aside, which in most designs nowadays, is a matter of moving a small planting tray aside, as easy as relocating a potted plant. So again, why are we so afraid of living roofs? The benefits of a living roof can be many, including: reduced energy use, reduced carbon footprint, increased sustainability, decreased storm water run-off, increased landscaped area for insects and birds, etc. A data center project I worked on in Switzerland required living roofs on the building in one of the Cantons (county) so that everyone’s view remained like the green grasslands it once was.

So, I challenge you to make your buildings as sustainable and beautiful as they can be, looking at all of the costs of the system, and consider a planted roof; you’ll love the view of it.

On a separate note, why have so few people looked at Canada to build a data center? The western half and other areas of Canada have very low cost power ($.02-.04/kWh), the power is almost all hydro, so low to no carbon, energy use would be lower as it is cooler, water is pretty ubiquitous so generally no shortages of water, land can be very low cost, fiber does exit major cities directly to Europe, US and Asia, people are friendly, etc. Perhaps the taxes are not ideal, but economic development incentives are attractive in some regions. Another thing to think about when building your next data center or other project.

When I started in the data center world, it was in the mid- to late-90’s, and I was excited by all of the unique equipment and great need for security and reliability. I was drawn to the UPSs and the business critical functions that needed them. While this was a learning phase for me, I was considering how we could reduce energy use while providing the desired level of availability. At the time, there were few options.

I remember asking crazy questions of our IT folks, like “can we shut off the servers when they are not needed and just turn them back on when needed?”, and “boy, these UPSs sure use a lot of power in addition to the plug loads, are there any better solutions?” Then, my next foray took me to Exodus Communications, the big powerhouse of data centers, where we built billions of dollars of them in short order, with all of their shiny new UPSs, generators and the like. In design meetings, we would always talk about how to improve reliability and my energy efficiency options were like a little squeak squelched by the ever-powering quest for more reliability, even if the energy efficiency options improved it or did not reduce reliability.

Now, fast forward 10 years, and we’ve learned quite well how to supply high-availability. Better yet, the applications, other software and hardware has come a long way in allowing us many options to reduce energy use while improving availability. Yes, on the design side of data centers, we still don’t marry these two paths very well into one holistic solution to availability AND cost, i.e. energy efficiency. For example, when building new data center capacity, we often think about the availability of the data center in its site selection and design but not the actual product that will be produced from the hardware and software inside. Most folks still think that if they want high availability, they need a Tier IV data center. Yet, in all of the data centers I’ve been involved in over the last 10 years (hundreds), in every case, we could provide two Tier II+ to III+ data centers for the same price as one Tier IV. And guess what, put those two data centers in geographically diverse locations and you have a net higher availability than the one Tier IV and at about the same or lower capital cost and a MUCH lower operating expense. In one example, we built two Tier III data centers for about half the price of one Tier IV data center. So, let’s look a the numbers: 4 times the capacity, a combined availability of one extra nine and an operating expense of about 25% higher than the one Tier IV data center. Now realize that every data center will fail in it’s life (us humans are imperfect and the equipment does fail), or that it will be impacted by a natural or human-caused event, so two geographically diverse data centers will have a low coincident chance of a simultaneous failure compared to the one data center. So, save money in construction and operations and gain higher availability; seems like a good option for many data center applications, although I know that now all apps and systems are capable of easy and automatic failover between sites. Now take your data center load across three or more data centers, with each one capable of failing over to a group of them, with only minor overhead of hardware to accommodate one off-line data center and you have a much better solution. This is the solution where thinking bigger in smaller building blocks has an advantage over thinking of only one option at a time.

Why is it that so many people think energy efficiency needs to cost more or that they question the results of good design? Is it that they question what they are doing themselves? Take this case in point: at the last Critical Facilities Round Table (CFRT) in Sacramento at the ADC data center, many folks came, looked at the design, listened, and questioned the results. I LOVE the fact that people are questioning the results and asking good questions…love it! Without it good debate we do not have dialog or can advance a design or issue. It is essential to our learning process, especially for those with many past experiences. It is good to question. But what I question is why when there are so many good examples of good design, so many people still achieve conventional results.

For example, the ADC site achieves pre-certified LEED Plantium and a calculated PUE of 1.12 in fairly hot Sacramento all for about the same cost as a standard design data center. Sure, it may not meet the needs of everyone, because data center needs do vary, but the site should be used as a guide to design ideas and of what can be achieved.

Let’s take another example: a site that I’ve been working on with Rumsey Engineers has a calculated PUE of 1.08. Sure the design is only 50% complete, but that is the lion’s share of what drives the PUE. Now the site is roughly a Tier III design, so not super reliable, and it is in a fairly favorable climate although it still gets to 90F in the summer. The low-PUE design is being achieved with what I call “good-smart design”, meaning without costly embellishments, no wacky technology plays, nothing unconventional, and estimated to cost not much more than we were spending to build a greenfield data center on a per kW of IT load at Yahoo–in other words, low cost! Now here’s the wacky part: it is designed to support rack loads up to 150 kW…per rack! Yes, that is an outrageous number, at least at this time, but it is what is the forecasted need for this uncommon compute need, but that is beside the point. The point is, that for roughly the same cost as a low-cost data center, at a moderate/average availability design, this site will use about 1/4 of the infrastructure energy to support the IT load as a “good” data center design of today. And since energy is the highest portion of a 10-year NPV of a data center, this is where these efficiency savings from good-smart design add up to millions of dollars saved annually, and that my friend, is just one of the many examples I can provide as to how energy efficiency does not need to cost more, it costs less and continues to give and give.

Use these examples to open up your thinking about what a data center should be and can be.