HealthITGuy's Blog

The evolving world of healthcare IT!

About HealthITGuy

Michael Heil has been working in the healthcare IT field since 1995.  His approach to IT has been to look at what makes logical sense, use deduction to determine the answer, maximize your resources and allow your staff to be creative and challenged.VMware vExpert

  • Member of HIMSS
  • Cisco Unified Computing Technical Advisory Board Member
  • VMWare vExpert 2009, 2010
  • VMWare Users Group Leader, Greensboro, NC
  • Bachelor of Arts from Michigan State University
  • Masters of Science from Michigan State University 


  1. Hi,

    I am the editor of and am working on a story about some benchmark testing done by The Tolly Group on the Cisco UCS that claims throughput to be much lower than Cisco promises. Tolly says that while the UCS chassis can house 8 server blades, each of which can have its own 10GbE converged network adapter, the maximum number of active uplinks to the top-of-rack switch is 4. Thus, even though the blades have a theoretical aggregate capacity of 80Gbps, they all have to communicate to the top-of-rack UCS 6120 XP Fabric Interconnect (switch) via the maximum of four 10GbE uplink ports of the UCS 2104 Fabric Extender that plugs into the UCS chassis.

    I am wonder Tolly’s desciption of UCS has being “oversubscribed” is worthwhile considering his set up may not be a real-world use scenario.

    Would you be able to offer some perspective on this?


    Comment by Rivka Little | February 25, 2010 | Reply

    • Rivka,
      Sorry for such a delay in replying, I had an injury and then surgery on my collar bone (xray on the blog :).
      I have read a bit about the Tolly report and gone through the UCS configuration used for their testing. My summary thoughts are Tolly did NOT use a real world configuration of UCS. They used a single 6120, which is like installing an EMC Clariion SAN Array with only 1 Storage Processor. Not only are you looking at half the bandwidth or capacity of a UCS configuration you are ignoring the redundant side of the configuration. Again you would never only configure or use a single Storage Processor on a SAN array, and stay employed very long.
      When UCS is installed as a normal (real world) fashion there is no over-subscription if you so choose. Meaning I can give a single blade chassis 40 GB of bandwidth to each fabric extender for a total of 80 GB for the chassis (i.e., 10 GB per blade) but I have the choice of giving that chassis a total of 40 GB or even 20 GB, depending on the expected workload. A key point being the admin has a choice based on the real world need.
      Thanks for the question.

      Comment by healthitguy | March 17, 2010 | Reply

  2. I am so sorry I didn’t see this response earlier. I hope that you have recovered smoothly from your surgery.

    Thanks for the response, it was very helpful.


    Comment by Rivka Little | June 25, 2010 | Reply

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 )

Google photo

You are commenting using your Google 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: