Gartner on Extreme Data Centers

One of the best sessions I went to at the Gartner Data Center conference last week, was entitled Extreme Data Centers – Attaining Massive scalability – in the minimum space at the lowest cost.

The talk was given by David Cappuccio who is a managing vice president and chief of research for the Infrastructure teams with Gartner, responsible for research in data center futures, servers, power/cooling, green IT, enterprise management and IT operations.

Dell's Modular Data center, mentioned in David's talk, a few days before it went live at Tier 5 in Australia.

Here are my notes from the talk:

The Extreme Data Center definition

  • Designed for efficiency first
  • Designed for optimal performance per kilowatt and or per square foot
  • They leverage new design principles to attain the biggest benefit

An example he then gave was supporting 2,107 servers, 18,300 images and 19 petabytes of storage in 2,600 sq. feet of IT space

New data centers are designed around efficiency

  • In power utilization
  • In space allocation
  • In capital expenditures

Three ways to solve your problem

  • Build your own datacenter from the ground up — greatest control but most expensive
  • Retrofit what you have to extend its life – greatest potential risk but least expensive
  • Use modular ideas to build and expand later
    • Reduce capital upfront costs
    • Simple growth when needed
    • Can use existing land or building

Emerging Design trends

  • Build small, build often
  • Build for density
  • Scale vertically and then horizontally
  • Build and rebuild pods (or sections of your data center)
  • Build density zones (group your systems by how dense they are – high, medium or low– and then match the power and cooling at the zone level.  Density is usually based on the workload mix)
  • Consider multi tiered designs (all apps aren’t created equal)
  • Use free air and reuse heat
  • Design for the unknown

Modular Designs for sustained growth/The evolution of pre-built solutions

  • David felt this approach was a really good idea and made sense
  • He felt the drawback was that there weren’t any reference accounts: he mentioned HP at Purdue and “some company down in Australia” (which dear readers is the Dell MDC down at Tier5, pictured above).
  • He cited Azure as the poster child for containers
  • Besides citing Sun’s “Black Box” as the granddaddy of all these, the pre-built solutions he mentioned were:
    • HP Flexible data center
    • IBM scalable modular datacenter
    • I/O anywhere
    • Dell Modular Data center

Energy consumption and efficiency

  • PUE, DCiE are defacto standards – use them
  • But PUE is not the goal – it’s the beginning
  • PPE: performance and capacity per kilowatt are key

Three examples of some cool new designs

  • Yahoo Computing coop: outside air cooled, minimal fans no chillers and a PUE of 1.08
  • Microsoft containers:  8×40 feet versions, 8-12 weeks for delivery, 2K servers per container
  • Net App: Slab-based w/overhead air design: first energy star rated data center, 25 megawatts

Stay tuned for more

The extreme data center space is an “extremely” hot one.  Watch this space to learn more about how Dell plays here going forward. 🙂

Pau for now…

Extra-credit reading:

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: