Cisco Blogs


Cisco Blog > Data Center and Cloud

The Napkin Dialogues: FCoE vs. QCN

July 21, 2011
at 2:49 pm PST

While at Cisco Live I had the pleasure of meeting several people who were curious about Multihop FCoE but had the unfortunate experience of getting too much misinformation from several sources (yes, including some of Cisco’s competition, but even some partners!). Some had already seen my article on FCoE and TRILL and wanted to know if I could help explain the relationship between FCoE and QCN (Quantized Congestion Notification), one of the documents in the IEEE DCB standard revision.

Even though we have a very good, short white paper on the subject, this is one of those subjects that as soon as people ask about it we break out the white boarding, or in the case of being at Cisco Live, the napkins. There are just some things that pictures help explain better.

Because of this, I’m going to try something different with this blog. It may work, or I may fall flat on my face; I suppose we shall find out.

Thing is, when I tried to write down this blog in the form of a conversation -- because it seems people are liking that format -- the blog grew longer and longer as I tried to explain these concepts in plain English. Creating the accompanying drawings made it even longer, so I’m going to jump in with both feet and try something radically different, something I haven’t seen in a technical blog before. As always, your mileage may vary.

So, without further ado…











Update: I would like to thank Erik Smith of EMC for pointing out that an earlier version of this blog had a pretty significant inaccuracy. In my attempt to simplify the process for the sake of writing this blog, I unintentionally conflated some of the FC operations and, well, no other way to say it. What I wrote was wrong (I hate being wrong! Especially in public!). I’ve since changed the graphics to represent the correct technology and hopefully have eliminated potential sources of confusion. - J

Tags: , , , ,

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.

26 Comments.


  1. LOVE the comic style napkin pitch… Great explanation. Thank you!

       0 likes

    • J Metz

      Thank you for taking the time to comment! It’s really good to know that an unusual approach can resonate. :)

         0 likes

  2. BRILLIANT!! I love Cisco, but we tend to be VERY long-winded. This format and explanation was perfect. Very easy to read and easy to digest. Thank you for taking the time!

       0 likes

  3. Awesome explanation…. How I wish all our AS Education material was so beautifully done, fresh and lucidly done!
    Looking forward to see you do this with a lot of other technologies! Keep Going!

       0 likes

  4. Just wanted to say thanks, I’m by no means a networking expert and wish there were more explanations out there as straightforward (and amusing) as this.

       0 likes

  5. Great job J..thanks a lot!

       0 likes

  6. Awesome!!! :-)

       0 likes

  7. Hi J Michel,

    I could laugh at each image and it is so easy to understand it.

    Would QCN be more helpful if in case of FCF switch in the middle, you would have a ‘dumb’ DCE/CEE/DCB switch, then the MAC source would be the one form the first FCF switch and it would be able to pause (without annoying the Host)?

    Thanks.

       0 likes

    • J Metz

      Thanks for the great compliment. :)

      To answer your question, you have to remember that FCoE networks *are* FC networks. That means that generally speaking, it’s often a good idea keep the same design principles as with traditional FC networks. Most multi-switch designs are edge-core, or edge-core-edge. That means that the number of switches you have in a FC network are usually 1-3 (at most).

      The average Network span where QCN will start to have advantages will be a minimum of *3* or *4* for a given traffic path. After all, if you have to hosts connected to one switch, QCN won’t bring you anything. PAUSE handles the link-by-link elements otherwise. If you start putting 3, 4, or 5 switches in the mix, you’re going to be moving farther and farther away (literally!) from FC best practices design.

         0 likes

  8. Very good job J. We read it together with Remi(22 months old) and I swear I saw a sparkle in his eyes. We are looking forward to read the next one.Thanks

       0 likes

  9. Very cool & well done!

       0 likes

  10. Great format, wish we had more like this!

       0 likes

  11. Nice creative approach J. Now, can we get that with animation and audio? :>)……. Actually, quite effective just as it is in comic strip format. Looks better than any napkin diagrams I’ve seen. Well done.

       0 likes

  12. Like it :). Can you do the rest of FC-BB-5?

       0 likes

  13. very nice… love the illustration!

       0 likes

  14. Hi J, great format!

       0 likes

  15. We need this for all our presentations! You rock dood!

       0 likes

  16. J.

    Nice job. The images really are worth 1000 words.

    Now why isn’t FCF functions applied at the edge and Ethernet in the middle not good enough? :)

    Mike

       0 likes

  17. Love the graphical approach. It’s easy and fast to digest and will likely be more memorable in the end than a regular post. Great job!

       0 likes

  18. AWESOME break down – THANK YOU for taking time to do this for the community!!!!

       0 likes

  19. Yay, great! Congratulations. :) You Cisco folks are awesome.

    Cheers..
    htto://www.kolkataways.com

       0 likes

  20. Well done Cisco! We use Cisco technology, but what makes Cisco special is the people it employees and the value they add. Well Done.

       0 likes

  21. Cleary graphical approach. very very good technology by Cisco.

       0 likes