Cisco Blogs


Cisco Blog > Inside Cisco IT

Would You, Could You, In a Box?

September 28, 2011
at 10:00 am PST

If you saw any coverage of the recent opening of Cisco’s Data Center in North Carolina, you likely noticed that it features two distinct server environments – an in-building facility and a containerized one.  Having them side-by-side naturally raises the question when is a container a good alternative to a brick-and-mortar installation?

I researched that topic a few years ago when I was part of Cisco IT’s Data Center design team, as we explored different approaches to address the company’s computing needs.  I found eight great reasons to consider a containerized Data Center, and three potential reasons not to:

  • Speed. Containers offer a way to quickly bring new Data Center infrastructure online faster than a full-fledged construction project.
  • Deferring capital costs. Containers allow for the incremental addition of Data Center capacity, allowing you to expand as demand warrants rather than constructing a large facility and taking years to fill it.
  • Space. If you need Data Center capacity but don’t have office space to spare, containers can be deployed in a parking lot or even on a sturdy enough roof.  Containers can often be stacked, letting you put even more infrastructure in the same footprint.
  • Risk avoidance. Expanding an in-building Data Center can mean dust, noise and downtime risk for hardware in the existing server environment.  Adding a container in the parking lot is less disruptive.
  • Energy efficiency.  Container PUE ratings are comparable to the latest and greatest in-building Data Centers.  Rather than painstakingly upgrading a legacy Data Center for greater energy efficiency, consider deploying containers and migrating your applications and functionality into them.  You can then repurpose or upgrade the legacy space, as appropriate.
  • Temporary capacity. Disaster responders and military groups use containerized Data Centers to bring computing and communication capabilities into the field – retrieving them once their tasks are done.  Data Center containers make an interesting alternative to a co-location facility for the same reasons.  Deploy them where needed, remove them when you’re done and – unlike a leased facility – you still own the container at the end of the day.
  • Modularity. Designing Data Center infrastructure in repeatable increments is a good practice.  Among other things, it allows you to gradually introduce new technology into the facility with minimal disruption.  Containers are an easy way to accomplish modularity.
  • Integrated computing. Containerized Data Centers are about more than just the box.  They can include integrated hardware and sophisticated environmental monitoring, resulting in a server environment with a lot of powerful technology built in.

With all of these advantages, why isn’t everyone using containers for their computing needs?  As useful as they can be, containers don’t fit all circumstances.  Consider:

  • Physical security. A container in a parking lot doesn’t have the same physical protections as an in-building Data Center.  (This can be mitigated by installing the container inside a hard-walled structure.)
  • Non-standard hardware. Data Center containers involve tight quarters.  This maximizes their use of space but doesn’t easily accommodate unconventional hardware.
  • One size fits most. A container’s rack space and power are what they are.  If their increments don’t align with your needs, you might end up with either unused capacity or not enough.  This can obviously be a challenge in conventional Data Centers, too, although it’s often easier to alter how much physical infrastructure is provided, especially in a large Data Center.

For information about Cisco’s Containerized Data Center in particular, visit www.cisco.com/go/cdc.

Tags: , , , , ,

Comments Are Closed