Routing IPv6
When planning for an IPv6 implementation in any sort of non-trivially-sized environment, one important step is to choose a routing protocol. Of which, we have a few choices. Today - let’s talk about our favorite “Interior Gateway Protocols” (IGPs). Note that this will be from an admittedly high level.
The goal of any IGP is to exchange routes (talking IPv6 here, mind you!) and thus establish a hopefully consistent view of the network and to maintain universal reachability within the network in question. There are many routes (pardon the pun) to this goal, and various routing protocols choose different methods to attain it. We could spend lots of time talking “distance vector” vs “link state”, but for the most part (today) we’ll deal with just link state options (who wants to hear about RIPng anyway? Oh, and we will touch EIGRP briefly - but that is more than just a DV protocol).
When choosing between IS-IS, OSPF and EIGRP there are a few questions to ask yourself. First off - WRT EIGRP - if you are an all Cisco shop, this isn’t a bad choice. EIGRP has a fair number of advanced (or “hybrid”) facets that really do make it quite attractive. If you are not an all Cisco shop, well - you have one less option to “worry” about. Maybe more on EIGRP some other day, but don’t feel bad - we will definitely be talking about EIGRP long before we spend any real blog-time on RIPng.
So, back to those Link State protocols - OSPF and IS-IS. Or, to be a bit more technically accurate, OSPFv3 and Integrated IS-IS. From a pragmatic perspective, the real answer for most environments is to choose whichever you have the most operational familiarity with. That is, all other things being equal - if you use OSPF (v2) for your IPv4 routing you would probably be best served by using OSPF (v3) for your IPv6 routing, and if you are more familiar with IS-IS then there you go. However, in some cases not all other things are equal. Maybe running a single, multi-protocol “integrated” routing protocol is desired. Or, maybe, maintaining segregated protocols is attractive. Those are two divergent ideas and I don’t know your network, or the goals you have for your network, so I can’t really help yo with that decision - atleast not in this forum! (It is a much longer, more interactive, discussion).
A quick list of IS-IS resources:
(Note - I am listing IS-IS first not because it is a superior IGP (although some do feel that way, and I am not saying they are wrong), but because it has had an update (or 6!) in the last couple days)
Several very recent RFCs: Routing IPv6 with IS-IS (finally ratifying something that has been well-supported for years!), ISIS TE Extensions, IS-IS + Graceful Restart, ISIS + GMPLS, IS-IS Authentication and IS-IS 3way handshake (PtP links) … all from the fine folks at the ISIS Working Group (although the site is terribly outdated (and even more so now!)).
And a couple of quick OSPFv3 resources:
OSPFv3 has had some updates of it’s own in the last few months … OSPFv3 (updated base spec), OSPFv3 TE Extensions, OSPFv3 Graceful Restart … likewise from the fine folks at the OSPF Working Group (and yes, this site is also a bit outdated)
Hope that helps, and see you online … or in class!
/TJ


October 6th, 2008 at 1:13 pm
TJ:
Which versions can route IPv4 and IPv6 in a *single* instance of the routing protocol? ISIS and OSPFv3?
The IETF OSPF group is working on adding IPv4 support to OSPFv3:
http://tools.ietf.org/html/draft-ietf-ospf-af-alt-06
I’ve seen this in action in 2006 with beta code for Cisco IOS in operation for the Army’s WIN-T routing backbone over JTRS radios. The draft for OSPFv3 that will carry IPv4 addresses is expired, but it is an active work item for the OSPF group
Also of interest, IPv6 routing for wireless networks is currently being standardized as extensions to OSPF for Mobile Adhoc Networks:
http://www.ietf.org/internet-drafts/draft-ietf-ospf-manet-or-01.txt
October 14th, 2008 at 12:03 pm
Dave - Currently just ISIS … more on OSPFv3, as you mentioned, in the future ;).
/TJ