This is additional coverage of a piece over at Data Center Knowledge about the adoption and competition of modular data centers heating up. Over the past 6-7 years, as long as we have been in the modular space, we identified many issues with the options, and the biggest issues had little to do with the technology.
The fact that there is M&A activity with Schneider/AST and a pending IPO from IO bode well for putting modular options in the public arena, little has been done to look at the (addressable) challenges that still exist with ANY of the vendors:
1. Focus on product vs. solution. If you talk to the vendors as we have done for 7 years, the positioning is that these are ‘solutions’. They are not, they are products. A solution is a sign & drive experience, not a buy and then figure out where we can put these experience. IO arguably has the most seamless solution out there with their IO anywhere product, assuming you like Singapore, NJ or AZ as data center locations.
2. No place to put them. When the modular products were introduced there was a lot of attention given to the ruggedness of the containers. They could sit outside in all sorts of climates. That’s great, only people are the ones who make these work so while the containers were rugged, the technicians didn’t like 115 degree heat in direct sunlight, or a Northeaster blowing rain sideways and being in a big box with a lot of electricity coming to it. Even NextFort (we would consider them a hybrid design) enclosed their offering which is largely built with the same materials as a Home Depot. They too realized that people are not as rugged as a product and enclosed their data center. There still is not a modular centric facility out there to support a solution. There are ‘architectures’ and ‘approaches’ but show me a building where we can roll a modular data center that we buy in, plug it in, commission it and move kit in. We want to see one that’s not on paper, we have those too already.
3. Marketed as a specific use case solution vs. platform. We don’t know how many data centers you have been in, but We have been in over 100 and for the most part every single raised floor data center looks the same. If pushed for a number we would say 95% are like that. The biggest difference at the facility is the sizes and shapes of the rooms. White walls (usually) white or off white 24″ x 24″ floor tiles, humming and or blowing sounds from the air conditioners and power distribution equipment and/or computer racks full of servers with fans in them. A container is simply another form factor to put in racks full of servers with fans in them. The biggest difference in data centers is between the airport and the front door, and that’s the stuff that is far more important than look and feel.
4. Lack of (referenceable) information. It’s a royal pain in the ass to find any data to help drive a decision on modular. For a few years no one wanted to say how many they had deployed because the answer was none, one or maybe two at best. The customers didn’t want the notoriety because then they might be called ‘out of their minds’ for using the technology that was not deployed widely and seen as risky. That stink has still not washed completely off.
So here is why we believe (and have believed for 7 years) modular makes sense:
1. They are products, not solutions. You are not locked into a particular vendor’s building, design, density, (poor) efficiency, or lack of support. You want a private cloud? Buy a container, it gives you 20-30 racks that you can run 750Kw to and that you can lock the door of the unit and the door of the container to. Your infrastructure, your container, your control. Blunt Hammer even has a design and a financial model done.
2. You can put them anywhere. Here is where the data center industry can really get interesting… I worked on a business plan for most of 2013 for investors who wanted to see what the opportunity was in modular. In detail. The findings blew many notions we had out of the water. One of the biggest was in costs. We found we could deploy a data center facility for ~$3M per megawatt. We found that we could deliver 20MW in 40,000 square feet along with electrical distribution, DR bunkspace, and a nice office or two. $60M for 20MW and a site PUE of 1.3. I remember when $11M per MW for a used facility with a 2.8 PUE was a good deal. If I am a traditional facility, I am hoping that someone doesn’t implement this business model because no matter how low I drop my rent, my real estate and infrastructure costs can’t touch this (queue MC Hammer)… The other thing we knew but hadn’t proved was that for all of the retrofits of facilities today (we have looked at breweries, former malls, old KMarts, and office buildings) using a container was faster and 60% cheaper. I would rather have 10 new sites around the US all geographically diverse and synchronized over my own network that cost me less than 3 facilities with a single vendor in a 100 year old building with janitorial staff that can hit an EPO button while dusting and shut me down (this just happened a few weeks ago at a facility).
4. They make an awesome stable, exact same platform. Think about this – I can have a handful of sites around the US with MY STUFF (private cloud, streaming service, gaming apps, healthcare data) running and stored. Synchronized, mitigation from disaster built into geographic failover, attached to major peering points, inexpensive to operate. Same skillsets as a data center to run & support. Guess what, If I don’t love them then I recycle them in the next tech refresh or change to a vendor with a better mousetrap and I still cut my computing costs in half on better faster equipment for that period of time. No vendor lock past the end of the hardware lease, and I have a true common platform and piece of equipment for my operations staff to know. It’s the Southwest operations model for data centers, only instead of Boeing 737’s its a container.
So are we glad to see there is more attention in the space? Yes. What we really want is to deploy the next data center platform that is already designed, already modeled, and needs the right investors who understand utility scale.
Blunt Hammer’s vision is to help deliver the next data center platform in more places around the world and operate the utility for computing. The mission is to give the world a data center platform that fixes the issues of inefficiency, cost, and uniqueness that is holding back the ability to compute better.
Please contact us if you want to learn more about what has been done. We have the team, the designs, the locations, and interested customers in hand.
mark AT blunthammer.com