Monthly Archives: November 2013

The Network is the Computer – Who said it first?

I was going through some blogs I follow looking for topics to expand on, or refute entirely, and I took a look at the Databank Wizard’s latest entry talking about connectivity is back in vogue in the data center and cloud space. They also hit on an important consideration – connectivity – when looking at facilities.  Modular or not.

There are a lot of real estate people in the data center business. They build beautiful buildings, get off on cap rate compression, de-risk deals using bricks and mortar vs. the technology inside their buildings. The risk there is that the technology inside will change very very very quickly. In fact, faster than it takes a broker to get his post digger out of the trunk and put up a sign. I have worked for the real estate centric players before, and now I take their phone calls about how they de-risk a deal because there is a business model that would commoditize their real estate that can be implemented in under a year for $40M.

Now what makes it all sing in the facility, is the connectivity. Especially cloud.  It’s the pipes we use to drain lakes and oceans of data. There are cocktail straws and 48″ water mains – which one do you want to be connected to if you’re trying to move an ocean of data?

So I am glad to see that there is more discussion about connectivity again. Because I remember when Scott McNealy told me at an event at Harvard we were both speaking at in 1998, that ‘The Network is the Computer’. BAM! Still true 15 years later.

Tagged , , , ,

The Commodity Data Center IS Available

I recently read a piece from Nick Razey the CEO of Next Gen Data Centers over at Data Center Dynamics where he laid out a compelling case for what a commodity is and how is doesn’t apply to data centers. I think it does.

Nick’s premise is that the definition of a commodity is same product, same quality of said same product, and same price for same quality product. The nuance and chief differentiator is a spot market. In other words a commodity is the same thing for the same price and a spot market is the key differentiator. Except data centers. They’re not a commodity.

Well, why not?

He points out that there are different Tier levels, different densities, different connectivity options, different staff training levels, and different procedures and this makes data centers commodity proof because the quality varies so much. Since there is such a variance in quality it’s buyer beware and data centers are therefore not a commodity. Really?


The reason for Tier Levels, SSAE 16, FISMA, TIA/ANSI, and a whole host of other standards is to insure that all of the quality issues ARE mitigated and sameness of quality and operation are adhered to. In other words – we have the definition of sameness, AND the yardstick to measure it. What are operators afraid of? Is it something else besides standards? Nick makes the case that no one with any smarts goes for the cheapest data center option but if gasoline/petrol prices were cut in half, we would all jump to the new gas. Why? With how different the refineries are in their age, location, safety standards, and additives it makes little sense why the gasoline/petrol business is a commodity at all if we use the same logic and argument. I call ‘bullshit’.

Another issue he points out is that capacity provisioning capacity makes it almost impossible to offer a spot market. Capacity cannot be scaled up or scaled down to meet demand. Really? How do we manage this crazy world of scale? Design and planning are vital steps to any project because they eliminate big problems and let you run ‘what if’ scenarios. Scaling up or down is a non issue, in fact, in a recent study we did we found that operators are really good at managing vacancies around 28% to allow for said scale. And licensed power vs. used power is where you see the issues.

The next point is that operators seek long term contracts to guarantee return on investment. Not sure why that’s an issue. Any due diligence I have ever gone through includes a look at finances to insure that every deal they do is profitable, it’s not a good sign if an operator did 50 deals last quarter and by some measures is knocking it out of the park – when in reality all of them lost money. A good operator will ALWAYS mitigate risk. Customers buy risk – physical, financial, operational. Period. Don’t think so? Then why is there mire than one Tier level for data centers?

Nick makes a couple of excellent points about the barriers of entry being high and how new entrants have not changed the paradigms. I would agree that the barriers of a successful entry are high and I will make the point, based on personal experience, that the investors are the highest barrier, and as a result, are reluctant to change paradigms in spite of proof on how to do exactly that. It all comes back to risk, and the fact that there is more money than data center operators out there. I could write a whole other blog post about kissing frogs looking for money (but I won’t). Nick also makes the point that to buy in and get a seat at the table costs well over $10M and I will disagree with this statement entirely. Some of it’s the money/investment component and the other is there are cheaper ways to skin the cat. The model we did had a facility built for under $5M (per MW).

The next piece of the post nailed it – that cloud services were more like commodities than anything. I agree. He goes on to say that cloud services may be the next commodity, which I started to agree with until I zoomed back out and realized that you can’t have cloud without a data center and because of that and by association cloud will never be a commodity because the raw materials (data centers) cannot be commodities so the whole premise is flawed.

The net of it all is this – If you have $40M you can absolutely commoditize this business in under a year. We proved it. And Nick just helped prove it in a roundabout way as well.

We should disclose that Nick or Next Gen data centers are not Blunt Hammer clients or prospective clients, and that our findings and models are proprietary and real. Blunt Hammer is talking with two groups who are intrigued with what we’ve done and we’ll see if there isn’t an entrant with the right investor  that finally does break in and break the business. Who wants to buy some risk?

mark AT


Why Modular is STILL Not There…

I just saw a video posted by Schneider Electric about their modular data center – you can watch it here.

Having followed and been part of the modular data center movement since its inception, and still the only person (that I know of) who has NOT worked for a modular manufacturer and has sold a modular data center (I sold a Cirrascale FOREST container to NASA) I am still blown away by the fact that modular data centers are STILL not solutions. They are products. I don’t care what Gartner says. Especially for multi tenant applications.

Blunt Hammer worked over the past year to build a business that was the first of its kind – a standardized multi tenant data center platform that was modular. The venture ran out of money and the initial funding source opted out of the A Round, however when I see videos like the Schneider one I still shake my head. Why? Because the elephant in the room and the very obvious and totally crucial piece of the solution is overlooked or assumed to be in place – a site where the modular data center will go or can go. Every video, whitepaper, or presentation makes a colossal assumption that a customer has the perfect site so the modular products can just snap into place. It doesn’t work that way. I’d be happy to prove it, call me.

IO is probably the closest option to a holistic offering, if you want your outsourced data center to be in NJ or AZ or Singapore. The limiting factor for their expansion (and Dell, and HP, and Cirrascale, and Colt, and BladeRoom) is that NONE of these guys bring site selection to the ‘solution’. Thinking that a customer wants to fool around with site selection, due diligence, purchase & sale agreements, zoning, CCR’s, and the rest of it.   I can tell you from experience, they don’t. If there is no place to put a modular or prefab data center then you don’t have a solution, you have a product that will be part of a solution. If you don’t have a methodology to drive how you find the right site, it’s the wrong site. There are very few real estate people who understand site selection, and even fewer with a methodology they can produce, let alone follow.

When we started down the path of modeling a company to deliver a modular platform, we started with what you need to start with – site selection. Once we determined the area of the United States where our site selection criteria was met, we focused on counties and regions where the methodology held up. Once we figured out the county, then we looked in 3 different towns. Then we looked at 7 different sites in the town we wanted to be in. We met with Economic Development people, we met with State Senators, State Representatives, even the Governor to make sure that the social aspects of a project were ok and wouldn’t create headaches initially or over time.  Once we had the site, we looked at the technology, not the other way around. Why? Because the technology and the products BY THEIR VERY NATURE can go anywhere. So we picked the spot where we would put the technology and made sure it was the best place to put it.

The other piece of the puzzle was we needed to do this at scale. Why? Because our belief is that a single facility is not a data center company, two is the lowest level of suitable, and that to have at least 5, preferably 10 is the way to go. So we found 5 cities to expand into that had similar attributes for low levels of risk and the right components of critical infrastructure.

So we understand this modular/prefab world as well as anyone. We know what goes into a project, into site selection, into design, into the business of a data center. If you want to deliver a successful project using modular technology, contact us. We know what we’re doing, and would be happy to prove it.

Tagged , , ,