Personal Blog (scroll down)
Innovation—Pre-Engineered Data Centers
Innovation—Pre-Engineered Data Centers

Innovation—Pre-Engineered Data Centers

Funny, one of the things they don’t tell you as a Cloud service provider, is that the boundaries of the Cloud aren’t white, fluffy, and ambiguous.  Quite the opposite as a matter of fact.  They’re cold, rigid, and expansion can be expensive.

Certainly, from the consumers’ perspective, that’s exactly what the Cloud is.  But use a leaf blower to clear away the puff ‘n fluff, and you’ll find there’s something rather solid inside.  And, it costs money.

In a past role as a Software as a Service (SaaS) product manager, data center health, capacity, and durability were daily concerns.  On a micro level, tactical activities, day-to-day, stood atop a three-legged tripod: RAM, CPU, and disk capacity.

The macro level, strategically, concern focused on the data centers housing my systems.  The common chair legs here were Wattage (electrical service capacity consumed|available); HVAC (air conditioning capacity); physical space (can we add more servers and/or storage); and, backup generator/UPS capacity.

Eventually, we outgrew our own data center. We physically were out of space for new servers; backup systems were beyond capacity; and, our HVAC couldn’t handle the load in the middle of the summer.

Alarms tripped. Beepers buzzed.  Staff scrambled. Ugly.

The potential solutions available were ugly as well.  Most likely, we were going to undertake expansion of the data center; bumping-out into the rear parking lot; consume precious parking spaces; and with ugly cost.  And, there was no guarantee by the time it was completed, that capacity would be on track with our needs.

Ever have to discuss with your CFO, why you had (unwittingly) over/underbuilt your $15M data center, that took 18-36 months to complete?

There’s a better way!
Pre-Engineered Data Centers

Not quite two years ago, I had a meeting with Steve Jacobs regarding a new idea in data center development.  In short, deliver to the market a pre-engineered data center, that can be quickly deployed, cost effective, and support incremental growth.

Now, the idea is reality.  Steve’s the CEO of Velocity Data Centers, anchored in the Chelsea/Ann Arbor area (Michigan, USA).  He gave me a tour of their new prototype.

When I encountered Steve’s efforts, I immediately saw the value.  HP and some others have some very specialized ‘shipping container’ type solutions, but they’re not the most attractive long term alternatives.

The basic description of the unit I’ve included pictures of, can be found here.  He confirmed the unit is currently powered for 100kW, but can be easily bumped to 200kW if desired (in the power room, I saw there’s clearly room for a second set of Liebert units).

imageIn the data package room, there are 2 AC units (cold aisle) already placed, with room for 5 cabinets per row as configured (20 racks total, this design).

imageGroundwork for the design (foundation, slab, etc.) is conventional.  The building itself is a specially formulated 4-6” thick concrete.  The structure even has a ballistics rating and can withstand a Category 5 hurricane!

If you are not responsible for data center operations or management, this may be so much ho-hum.  If you are responsible for paying for one, here’s how it boils down…the business case.

At this point, I’d like to point out that I have no legal or monetary relationship to Velocity Data Centers.  I simply believe in their approach, a lot, and am happy to evangelize.

Consider, building out a 10,000 square foot data center can range between $10-15,000,000 dollars, and take 18-36 months to complete.  Depending on external developments along the way, you may find your needs have substantially changed.

Here’s what I really like about this solution:

  • Overall package (building and fenced area) foot print: 3,200 sq. feet.
  • Cost – Under $1,000,000 for a 100kW, 20 rack data center
  • Design—Everything pre-engineered as a package
  • Time—70 days for functional unit (this includes the learning curve for the prototype.  It will be interesting to see how quickly production units can be deployed)
  • Placement—Can be deployed anywhere, parking lots or vacant property, with suitable services.
  • Investment—Sized to meet your needs. Add additional units (collocated or dispersed) to meet increased demand, or for Tier 1+ needs.
  • No disruption to current operations until you’re ready to ‘flip the switch.’  Though life’s rarely that simple<g>.

At the end of the day, even though many may choose to transfer corporate services to the Cloud, somebody has to actually run the Cloud.  If you’re somebody, where time, money, and availability are all pressing concerns, this is an innovative approach you owe it to yourself to take a look at.

Thank you to Steve Jacobs, Velocity Data Centers, for the tour.

One comment

Comments are closed.

%d bloggers like this: