Dog food tastes nice!

Dog food tastes nice!

At a recent customer meeting I had the good fortune to see a wonderful presentation from a chap called Rich Gore who works in Cisco IT.

Rich’s team shares information with customers about the “Cisco on Cisco” approach, which is simply Cisco eating its own dogfood by using its own technology, like UCS, across its mission critical systems.

When I was at VMware there was a similar, successful approach to consuming one’s own dogfood where critical systems like the order-to-cash systems, all Oracle, were run in production on VMware. I wrote about that on VIOPS:

In the customer meeting, there was one incredibly powerful slide that shouted out to me and everyone else when Rich showed a slide on the positive impact that Cisco’s Unified Computing System can have on your data center. Here’s the picture, courtesy of Rich (thanks, Rich!)

UCS Cabling

UCS Cabling

Why is this slide incredibly powerful?  Consider the following points:

  • For the data center engineers who suffer from OCCT (Obsessive Compulsive Cable Tidiness), the UCS photograph on the right is something they dream about – now it’s a reality!
  • Cables equate to ports, and ports cost money. How few cables (and therefore ports) are there with UCS?  How much money are you going to save?  Cabling and IO ports cost a lot more money than most people expect
  • Cables and ports don’t just mean money, they mean effort and complexity.  I think there must  be a formula for number of incidents per hour spent in the data center.  If you’ve ever worked in a data center, you’ll know what a noisy, cold, isolated, tiring and unpleasant place they are to work in.  People get tired, and mix that in with complex and difficult cabling it means mistakes are made: less physical cabling means less mistakes which costs less money.
  • The physical complexity of the left-most, “Ad hoc” picture is now moved into logical, software complexity which is much easier to manage and manipulate.  Instead of hundreds of cables and ports, you can use a GUI or, even better, script the whole thing from the comfort of your desk, or even at home!
  • You can see the UCS “wire once and walk away” approach in front of your eyes, it isn’t a marketing gimmick, it’s an operational reality.
  • Single network fabric and UCS provide amazing flexibility in the data center.  Behind that picture is a single Fibre Channel over Ethernet (FCOE) network; and a single network enables a lot greater flexibility about allocating services to resources.  With the UCS management tools, it provides the first step toward cloud computing in the data center.

Lastly, a very important point about UCS: it’s all about scale.  What you see in these pictures is one rack, which is showing just two UCS chassis in them.  Now imagine forty chassis across twenty racks (or fewer racks, as each chassis is only 6RU) – and imagine, in the opposite row, all the racks and cables required for a similar capacity without UCS.

Which row would you want to own and manage?  No brainer.  The UCS row.

Game Over

Game Over

Related posts:

  1. How real life became my worst nightmare
  2. Tradeoffs between scalability and performance in UCS
  3. Cisco 6100XP aint Switches like VMware ESX aint Linux
  4. VMware + Ubuntu + Ruby + REST + XML + Cisco UCS API
  5. Over-commiting your infrastructure for multi-tenant DR with Cisco UCS