PUE, DCiE, and other stuff that doesn't matter much...

The Gartner conference was one of their best attended events in 2011 proving yet again that the space is hot, appears to have a bright future, and like the other hot sectors before it, is trying to mature through the development of standardized measurements that will be embraced by vendors, users, and analysts and give owner-operators bragging rights and a new area to compete in. PUE, DCiE and other yardsticks are competing for the 'best' yardstick for efficiency.

Sorry to pee in your Wheaties folks - we already have one. Cash.

I have done enough deals to know that PUE matters before financial analysts get involved. Once they do it comes down to costs. Total costs. Lowest cost wins. Not every time, but show me a CFO who authorizes paying more for something than its worth as a rule.

I will also put a caveat in to say that this is more true of multi tenant facilities than single tenant facilities. Single tenant/purpose built data centers can do what they want so long as it makes financial sense and delivers on the business objectives for the company and THAT company only. Multi tenant facilities must be far more INCLUSIVE of a wider array of requirements to service the greatest number of clients and their desires.

Both Single and multi tenant data centers will have a mix of manufacturers, densities, layouts, loads and preferences. The multi tenant facility needs to factor in broad compliance, placement of densities, weight, and other highly variable nuances. Like Rick on Pawn Stars says - 'You never know what is going to walk through that door.'

Gartner Data Center Conference brain dump...

I just got back from 3 1/2 days in Las Vegas at the Gartner Data Center Conference, and will be writing about many of the things they presented, only in a lot more detail. In fact that was the #1 piece of feedback I heard was that the presentations are more sizzle than steak, and that they have to do a WHOLE lot more in providing the details about what they present.

Monday's keynote was pretty good and was given by Dave Cappuccio. It covered many stats and numbers that were fun to listen to, and I thought it was a decent way to ease into a conference. Some of the interesting nuggets I heard:

30,000,000,000 (billion) pieces of content were added to Facebook in the past month
Worldwide IP traffic will quadruple by 2015 - care to guess why?

Data Centers consume 100 times more electricity than the offices they support
The percentage of a college student's texting to phone call ratio is 98.4% to 1.6%
Within 4 years the bandwidth I/O per rack will increase 25x

Needless to say this is all pointing to growth - whether a company is prepared for it or not. There are some other topics I will be jumping into over the next few weeks as Gartner left a lot to be desired and discussed and in some cases were pretty off. So Stay tuned and lets see what I come up with once I shake the red eye jet lag...

Is innovation exclusively for single tenant datacenters?

I am a data center geek. I love to read about them, see who is doing what, learn about new stuff and approaches that are being implemented - all of it. What I have begun to notice is that the innovation that is happening - while totally awesome - will never fly in 95 out of 100 facilities in the world. I have begun to ask the question 'why' to other geeks and data center users.

Without question the number one response is - because when you run a single tenant facility you can take risks and do things that mitigate a common set of risks - and since one company pays, they can do what makes sense for their business alone.

Others include - 'because they have an innovation budget'; 'it's about controlling their destiny'; 'the penalties for an outage are not as severe as outside the single tenant walls'.

In my opinion I think it has to do with a combination of factors - most of which have been mentioned above with one exception - evolution. As companies move from a cabinet in the ghetto colo company, they evolve to require more choices and options, different configurations, and vendors who deal with companies the size they hope to become. Once they scale up to ~10MW (100,000 square feet) they evolve into looking at finish to suits and finally building their own. Why do I think this?

I saw it happen with Facebook. They started by leasing cabinets in a few key markets, scaling that, then taking cages, then taking suites, then floors of buildings and now are building their own. Their headcount for their data center operation grew to that of a mid size provider, and there does come a point where it is more cost effective to build your own than to have to keep moving like a shark looking for open suites, floors, or whitespace.

I have seen it happen in reverse with many banks too. They built their own and leased where they wanted, acquired with abandon, and then had an impressive footprint. Then they figured out - as many did - that the network was more important than real estate and so the consolidation began where they wanted to consolidate into fewer huge footprint super centers because it was more effective to run a few huge facilities vs. dozens of all different sizes.

So when I think about why does all of the innovation appear to happen in the huge facilities, I look at the evolution of Apple, Facebook, Google, ING, Citibank, Morgan Stanley, etc. and they all find a different path to the same location - a risk averse facility that makes economical sense. Innovation is tied to reducing risk or cost and when you are the only one who has to live with the decisions and their outcome, single tenant facilities will continue to out-innovate the multi-tenant facilities since they have an exponentially larger set of risks and requirements to service.

What do you think?

Google's Data Centers use 26 Megawatts - What's the fuss?

In the past 24 hours there has been a ton of press about Google's announcement that they use 26 million watts of electricity. Put another way, that is 26 Megawatts, and put another way, what is all the fuss about over 26 megwatts?

Is it that they ONLY use 26 Megwatts?
It it that Google is so efficient that they run this gigantic brand on such a small amount of electricity?

That has to be it, because folks, 26 megawatts in the data center business is a medium size operation.

Just for comparison's sake, Vantage data centers is building a campus twice that size in Santa Clara. Digital Realty Trust has a project in Dallas that is 4 TIMES that size to service clients. Bytegrid Holdings LLC has a 9.6 megawatt facility about one third the size of Google's entire operation. Just sayin'.

What I also have yet to review fully is the carbon footprint. It was mentioned in another, better,  article that they get 25% of their electricity from renewable sources. Not LOW CARBON sources, but renewable sources. The thing I have paid attention to for years is what I call the carbon chain - what is the carbon footprint from start to finish on a project, and then the impact on an ongoing operational basis. And I'm sorry buying a carbon credit doesn't count. That's like buying recycled paper with Monopoly money.

It mentions that Google builds their own facilities. What about the carbon it takes to manufacture the steel, the concrete, the copper, and other raw materials used in a facility? Then there is the transportation. How many trucks filled with diesel are trucking in heavy loads of new equipment that was produced from scratch and had to source the raw materials, and you get the picture.

Why not re-use a facility as a rule vs build new? Why not cap the radius of transport of newly manufactured goods to 100 miles or less? Why not look at the whole impact and subsequent carbon footprint and measure and improve that? Just sayin'...

Still, great numbers by Google.

And a side note to Noah Horowitz at the Natural Resources Defense Council who  '...cautioned that despite the advent of increasingly powerful and energy-efficient computing tools, electricity use at data centers was still rising, as every major corporation now relied on them. He said the figures did not include the electricity drawn by the personal computers, tablets and iPhones that use information from Google’s data centers.

“When we hit the Google search button,” Mr. Horowitz said, “it’s not for free.”

Trying to link energy usage of other corporations, personal computers, and iPhones to Google's numbers is like blaming a person's alcoholism on the fermentation process. It's a stretch at best. And what are you going to do to stop this reckless usage of the most efficient data center footprint on the planet, stop Googling? Didn't think so. Me neither.

Nothings free, but this is data that shows if we are going to search and want to be green about it, Go Green, Go Google.

Hybrid Cloud = Same Sh*t different Offering?

I have had a number of conversations the past two weeks about Cloud computing and what companies are asking for and what they are actually buying. My observation is this:

Companies want providers to offer public and private cloud elasticity, but only buy hybrid.

This is following another form of pretzel logic (yes I like Steely Dan. A lot.) which are the folks who say they need to be outside of THE blast zone. (What zone? For which type of blast?)

First one first - duh. Was there ever ANY other choice than they hybrid cloud? Not all data is meant for public consumption in spite of compliance rules trying to impose transparency (visibility) and so the notion of a public cloud was right out of Woodstock, man. The public cloud was exciting because it took the pressure off of IT departments to have to plan way ahead of stuff. If they needed capacity, they went out and got it someplace that had it and didn't have to go to a CFO and ask for $500K of gear to support the new marketing initiative or cover their *ss when they underestimated the website traffic when Justin Bieber played a benefit concert for Hurricane Irene survivors. It solved a lot of unmaterialized logistics and scoping problems and was branded as being insecure and we couldn't have that in the age of WikiLeaks and Anonymous hacking could we...

Then the pendulum of cloud swung back to private cloud. Well by my assessment private cloud was just another way to get billed for computer stuff by the hour. Private Cloud computing was becoming the No-tell Motel option and charging by the hour. If you wanted something quick and dirty and the movie titles not printed on the receipt then you could get your resources by the hour. In this case all the data and network and everything about the environment was kept out of plain sight and secured.

Then it's as if the blinking light went steady on us - we can have both things - the ability to not have to know in advance what we will need for resources AND the ability to keep sensitive stuff more secure.

Is it just me or isnt this what managed colocation and managed services have been doing for the past 20+ years?

Why are containerized modular data centers the new Black now?

What took you so long?

I was at Data Center Dynamics in Wasington DC on Tuesday and you may have thought it was Modular Dynamics instead. It seemed as if the data center world finally came around to what I have been seeing, studying, deploying, and writing about since 2008. That's not to infer that the modular solutions have gone mainstream yet, but the conceptualization has matured a lot as expected.

HP had a scale model of the EcoPod which was cool to see since up to that point I had only seen Powerpoint slides about it and have yet to see the real thing. PDI was showing off their modular solution as well which will be watched closely as they have some patent infringement allegations they are fending off. ActivePower had their PowerHouse scale model on display as well and given the coziness between HP and ActivePower in recently announced deployments I could easily see them sharing a booth at future trade events.

Bytegrid was the only data center owner operator who was able to speak to containers at the event - and did openly talk to several people about the good the bad and the stuff that trips you up on deployment and had specifics to back up the discussions. The three other owner operators there - GigaPark, QTS, and Powerloft weren't talking about supporting modular solutions.

Based on what I saw, acceptance has increased dramatically, demand is rising right along with it, and the Achilles heel that was there three years ago is there today still - being able to answer 'So where can I put one of these things?' with something other than 'Wherever you want'. ByteGrid is on to something and will be watched.

Rethinking the Value (and cost) of data

I have had a number of conversations recently with some hospitals about storage requirements, and specifically where to put more gear as their data centers are bursting at the seams, or in some cases becoming structurally weakened because of floor loads. What I figured out pretty quickly is that there is an issue that is fairly ubiquitous to these organizations, but an issue that I believe spills over into any firm who stores data - all data is treated the same and is costing firms millions each year to treat all their data the same.

Here is one example:

A hospital is looking for data center space to grow their storage footprint into as they have tapped out every watt of space they have in the hospital. That's not even the real issue. The real issue is that because they have so many racks of gear, and heavy storage arrays on floors of the hospital with floor loads insufficient to support the weight, that the hospital is experiencing structural issues. Floors are sagging under the weight of all of the data being stored. Literally.

Many vendors they have talked to are simply telling them that's unfortunate but they need to buy more storage. However, the vendor won't get the sale unless his uncle is in the construction business and can structurally retrofit higher floor loads before the new arrays show up. So what does the hospital do?

Hospitals see patients, maintain facilities, manage compliance conformity, and administer services. Their IT guys set up storage arrays, networking equipment, and manage the applications that run the business. They are not in construction and they are not data center experts and yet are the go to guys to figure all of this out.
The solution that we began to discuss - and that I wanted to share - is that there is a fundamental change that MUST take place in how they think about data - storing it, managing it, accessing it, and realizing it's NOT all the same. The other thing that had to happen was to look at the way their business operates (no pun intended) and structure things in a way that are based on how their business runs, and not how their data flows today, but how it needs to flow based on their business.

Here is a specific example -

New patients create alot of new records. New paperwork, new insurance information, new MRI's, new test results, new billing codes, new invoices, etc.

Existing patients have this data on file and their data footprint changes with an office visit, an MRI, a perscription, and all of the associated charting that needs to happen to document the recent activity and results.

What are the commonalities of the two different kinds of patients? They create data around events - visits, procedures, tests. What are events tied to? A date. Can the date field be used to assess the freshness of data, and another date identifying the next event indicate the likelihood of accessing that data? Hmmmm. Interesting thought stream just started.

Where we got to was that whether or not the patient was new or existing, their events drove the need to access data. A new patient would likely need to be seen again shortly and would need to have their records accessible whenever. An existing patient who came in to have a sudden sports injury looked at would probably need to have their data file accessible since there would be consults, MRI's/X-rays, referrals, etc. taking place rather immediately.

A patient who was on a 'check in once a year' cycle, doesn't need their data accessed 363 days a year on average. They come in, get looked at, maybe a test is required (but we know what it is ahead of time) so why is the 'active' patient's data sitting with a 'maintenance' patient's data in the same facility, withe the same cost, and with the same SLA (service level agreement) or DAA - data access agreement? Why would you pay for 363 days of something you don't need? I equate it to paying for a rental car in Seattle for 365 days, while I am only in Seattle two days a year - why would I do that?

The point here is that yes, they do need more storage. They can't put it where they usually do - in their data center - and need to find someplace else to expand to. But before they just go out and buy new arrays, new cabinets, new servers, and lease new space, they MUST look at things in a new way so that not only are they solving the immediate issue -'we need more storage' they are solving the fundamental issue - all data is NOT the same and does not need to be managed the same, or cost the same.

Thoughts?