Static routing with Catalyst 3750: and the winner is …

The Static routing with Catalyst 3750 post has generated a lot of good, creative ideas. Some of the proposed solutions were better than the others and some were simply not implementable (but nonetheless, had great creative potential :). Here is my list of the favorites:

A routing protocol: as a few of you have rightly pointed out, this is the best choice.

Aggressive Unidirectional Link Detection (UDLD): this is my second favorite, as it's a reliable link-level mechanism that will detect a break in the fiber cable … exactly the right tool for the job.

Object tracking and reliable static routes would also work. This was my initial solution, but I was worried about its support in Catalyst IOS images. In the meantime, one of the readers has noted that the reliable static routing (or at least the configuration command) works in IOS release 12.2(37)SE, so this might be a viable solution.
GRE tunneling with GRE keepalives probably has performance issues (I am not sure GRE is ASIC-switched on Catalyst 3750). The idea is to create a GRE tunnel between IP addresses on the primary fiber link. If the connectivity breaks, but the subnet remains available, the GRE keepalives will detect the failure.

Spanning tree will not work. It does not test the two-way connectivity and might actually create a loop (if I stop receiving spanning tree hellos, I might assume the link is connected to a workstation and OK to use).

Etherchannel and LACP would most likely fail. If the radio adapters work as switches, so you cannot establish an Etherchannel across the two links. I am also not sure that the PAgP or LACP would detect a unidirectional link when the carrier is present on both ends. Any experience?

Bidirectional Forwarding Detection (BFD) is not available on Catalyst 3750 at all (the ISR routers got it as late as IOS release 12.4(15)T). Furthermore, BFD (as implemented in IOS today) detects a routing protocol neighbor failure, not an IP next-hop failure, so you need to run a routing protocol first (in which case we wouldn't be discussing this scenario, as we would have changed the boss, as someone has suggested).

Edited on September 9th to clearly indicate what would or would not work

8 comments:

  1. Ivan! :) you did not specify which one is to be used in your scenario as you said in your previous "questioning" post! Is this for us also to guess again which one is the best among those choices given?! :)

    ReplyDelete
  2. About STP I don't really know why you wouldn't use that. If I understand that scenario, you have switches on both ends of the link. If you lost connectivity on fiber you won't get BPDUs on both ends of the link, so switches will choose link with next cost, so only thing you need to do is alter stp cost on ports which are backup.

    About etherchannel it probably won't work in that case (backup on a radio?), because I was trying to make etherchannel over two wireless Proxim Tsunami links and it wasn't working properly (i guess because of latency which breaks LACP).

    ReplyDelete
  3. @Anonymous: You're right. I've edited the post to be more clear.

    @Wrench Head of the Dirty Drunkards MC: STP won't work if the break in the fiber is unidirectional, thus one end receives BPDUs, the other one doesn't. Also, if you stop receiving BPDUs on the link, you don't block it, you assume it's safe to use.

    ReplyDelete
  4. At 2006 Networkers a presenter said that GRE on the 3750 is software switched and is unsupported and discouraged. Don't know about the 3750-E. FYI.

    ReplyDelete
  5. This is strange. I have many fiber mediaconverters (but mostly unused now because using gbics) and if you unplug one fiber (TX for example), whole link goes down.

    Next thing: you receive bpdus on TWO ports - one from radio, second from fiber. If you stop getting BPDUS from fiber, It will elect new root on radio port. It will also "see" MACs of other side on radio port, not fiber port. So it will send traffic trough radio, but it is true it won't block port of fiber.

    ReplyDelete
  6. @Wrench: not all media converters are as sophisticated as yours (unfortunately). I will simulate the STP scenario with Dynamips and post the results (I know what will happen, but then you'll have the chance to verify it :).

    On the LACP issue: LACP is supposed to work hop-by-hop, not across multiple bridges (oops, they are called switches these days).

    @anonymous: Thanks for the GRE tip, I suspected as much (high performance L3 forwarding in Cat3750 is ASIC-driven and ASICs cannot insert GRE header, just swap the L2 header).

    ReplyDelete
  7. BFD is available. 12.2(40)

    ReplyDelete
  8. Thanks for the tip :) Do you have it up and running? BTW, BFD is not listed in the 12.2(40)SE release notes.

    ReplyDelete

You don't have to log in to post a comment, but please do provide your real name/URL. Anonymous comments might get deleted.

Ivan Pepelnjak, CCIE#1354, is the chief technology advisor for NIL Data Communications. He has been designing and implementing large-scale data communications networks as well as teaching and writing books about advanced technologies since 1990. See his full profile, contact him or follow @ioshints on Twitter.