Ultimately, my explanation here was intended to be generic, to be deliberate in showing that there are different forwarding mechanisms for the types of traffic. Once we understand that FCoE isn’t restricted by the types of problems TRILL is supposed to solve, the question can then turn (and now it can) to the use cases you ask for.
Those use cases deserve their own focused post, however. Sadly I haven’t begun them yet, but will as soon as I can; I wanted to run this up the flagpole first to see if there was even an interest in going deeper.]]>
I’d like to see some references to design material that explains the design and use case. Hard data to put some meat on these bones.]]>
Thanks for taking the time to reply to my little post.
When referring to “we”, I was simply referring to b) “we” in the industry as its generally implemented.
Re: BTW1 I’ll have to take a look at your PDF file. Thanks for passing it along.
Re: BTW2 Thanks for raising the point about loops adding latency. You are, of course, absolutely correct and it is yet another reason why proper design for storage networks is paramount. One of the key elements of converging networks is that the basic assumptions for developing Ethernet networks are not always the same as the assumptions for storage network. Obviously, exploring the nature of proper design, using the proper technologies appropriate to the architecture, is beyond the scope of this blog article. It is, however, worthy of a much more focused piece.
As far as the history of FSPF, others more familiar than I with its story should probably address that question.]]>
Can you clarify on a couple of points, specifically, in the following paragraph, you use the word “We”.
Is “We” in the context of a) how Cisco implements within your products, or b) are you referring to “We” as in the industry implements or c) what is in approved or pending standards?
You also mention FSPF, care to comment on what FSPF is based upon or very similar to (caution, may open old wounds for some).
BTW1, came across this piece that looks back at what could be the next dimension to the FCoE/DCB/TRILL discussion which is PIM or protocol intermix mode.
BTW2, whether loops are good or bad, they can add latency which is not a good thing for data and storage networks. Likewise hops are great in beer, however they can add latency to networks, not to mention slowing down those who consume too much of it as well.
Greg Schulz – Author The Green and Virtual Data Center (CRC), Resilient Storage Networks (Elsevier) and coming summer 2011 Cloud and Virtual Data Storage Networking (CRC)