What Price Capability?

August 3, 2011 - 4 Comments

I realized a few years ago that all Data Center challenges can be solved with the sufficient application of money.

Need more computing capability?  Buy new hardware.  Struggling with hot spots?  Purchase supplemental cooling infrastructure.  Don’t have enough physical space?  Pay to expand the Data Center or lease additional space.

More performance means greater cost, though.  Some energy saving technologies buck that trend when compared to conventional facilities, but generally the more capability you want from a Data Center the more it will cost to build and operate.

I speak regularly with customers about Data Center design principles – the ones we follow at Cisco when designing our Data Centers and what qualities the people I’m talking to want from their particular server environments.  You can probably guess several of the principles – high availability, energy efficiency, scalability, etc.

The final Data Center design principle I invariably discuss is business value.  It’s the term I use for defining the relative importance of everything else on the list.  For instance, is uptime important enough to the company to warrant the cost of providing redundant air handlers and standby generators?   The more accurately someone knows the functionality and performance they need from their Data Center, the better it can be designed.  Overestimating leads to overbuilding and overspending; underestimating leads to a Data Center that likely won’t perform as desired.

See below for discussion of how various Data Center elements affect the cost of the facility.

In an effort to keep conversations fresh, Cisco Blogs closes comments after 60 days. Please visit the Cisco Blogs hub page for the latest content.


  1. Hmm… I sense a future post in the offing, asking folks to suggest odd/creative Data Center metrics to the industry. SODA CANS definitely goes on the list.

  2. Hi Douglas,

    Thanks for your reply…

    If the UK Government can officially come up with a scale for measuring the nations happiness, then I’m sure that we could come up with a suitable metric for such a requirement.

    You are right, it could be related to Superfluous Organic Dust Accumulation Causing Adhoc Network Spikes (SODA CANS).

  3. Hi Douglas,

    I agree that most problems can be solved by spending money, but as always it’s a risk reward argument. Can you get away without spending that money and if you can what can you save in the process?

    We provide specialist data centre cleaning to a large number of companies who contract because they see the risk/reward.

    My question is this…can you see a day when Cisco’s warranty would be dependent on such a service? If you buy a car to retain the warranty it needs to be serviced regularly, so if you buy hardware that’s attracting dust should the warranty be based on proper maintenance?

    It all comes down to the risk reward argument…

    • Hi, Jonathan –

      Hardware manufacturers frequently call out suitable environmental conditions for their systems – temperature ranges, clearances for airflow, grounding instructions, etc. I haven’t ever seen a warranty requirement that the environment in which a computing device is operated be cleaned on a regular basis, though.

      It’s a creative suggestion, although I’m not sure how practical. Would the warranty specify maximum particulate counts for the Data Center? An acceptable ratio of half-open soda cans per square foot? Something else?