It can be useful to have some basic information about how the various First Hop Routing Protocols (FHRP’s) work. When I read that the first release of Cisco OTV (Overlay Transport Virtualization) requires manual FHRP filtering, I started wondering how I would do it. The Cisco documents talk about MAC address-based filters. Ok, to do that, you need some basic info. Hence this quick note with what I found.
Murphy’s Law applied: I looked for info for a while, found most of it, and then found a web page listing all the information (URL is at the end of this article). And I probably should have started with Wikipedia (it’s been pretty useful as a technical reference lately!) My hope is that by repeating the info in one place it’ll be helpful.
OTV tip: The Cisco documents also mention that your FHRP gateway should not be on the OTV devices, i.e. the SVI (interface VLAN) must not be in a VLAN transported by OTV. One guess is that this is because manual or automatic FHRP hello/advertisement filtering won’t work on such an interface. The workaround if your Aggregation layer is the datacenter Layer 3 switch AND the OTV edge device is to do OTV in a separate VDC.
The promised info:
Hello/advertisement: Sent to the general all-router multicast IP 220.127.116.11, UDP port 1985
Virtual MAC (VMAC) used: 0000.0c07.acXX, XX = HSRP group number in hex
Hello/advertisement: Sent to dedicated multicast IP 18.104.22.168, IP sub-protocol 112
VMAC: 00-00-5E-00-01-XX, XX = the Virtual Router IDentifier (VRID),
Hello/advertisement: IP multicast address 22.214.171.124, UDP 3222
All of the above turned out to be at the following URL, with a bit more good summary info about how the protocols work:
4 responses to “First Hop Routing Protocol (FHRP) Info”
Has this changed? From what I had read, OTV did not require manual filtering?
The first release of OTV doesn’t do some things. I will shortly be posting a presentation I did yesterday on OTV for CMUG (1/19/11), containing a list of the things I know OTV lacks / might have in the future. The two main ones are: manual filtering for the FHRP, and adjacency server functionality for operation over a unicast WAN. These are supposed to be in the forthcoming next NX-OS code release.
I intend to post a follow-up blog to this one, because in presenting, it occurred to me that one probably wants to not only block the FHRP itself between sites, but also block any ARP directed towards the Virtual IP used by the FHRP. I intend to revisit a Networkers presentation from last Summer and review the ACL I noted there, and will post a blog supplementing the above material to make sure readers get the full story, if that is necessary.
It turns out the Cisco Networkers example is pretty good for HSRP, but doesn’t block ARP to the VIP cross-site. The local VIP will probably win the race, but one might want to be sure. (Good question someone asked yesterday during my presentation!) I’ll leave that part as an exercise for the reader — the Nexus does support ARP ACL’s.
See the new posting at [url]http://www.netcraftsmen.net/resources/blogs/cisco-overlay-transport-virtualization-otv.html[/url].