Unfortunately I don’t get to do much v6 at my job yet but I still like to stay updated on
what is happening. Do you run any v6 in your network? If so, do you run it native or tunneled
or something like 6PE?
Here are some interesting sources for operation of v6.
draft-matthews-v6ops-design-guidelines-01
This document discusses if IPv4 and IPv6 traffic
should be mixed on the same interface or should different interfaces be used? Should
link local or global addressing be used for routing? Should v6 routes be transferred
over v4 in BGP sessions?
draft-ietf-v6ops-enterprise-incremental-ipv6-01
This document is for deploying v6 in an enterprise network. Things like security policy,
addressing plan and IPv6 myths are brought up.
This document is purely about the advantages and disadvantages of only running link local
addresses.
Also, don’t miss out on information that is freely available at Cisco Live. Here are
some interesting sessions on IPv6 from Melbourne.
BRKRST-2301 – Enterprise IPv6 Deployment (2013 Melbourne)
BRKRST-1069 – Understanding IPv6 (2013 Melbourne)
ITMGEN-1313 – Preparing for IPv6 in the Enterprise (2013 Melbourne)
BRKRST-2311 – IPv6 Planning, Deployment and Troubleshooting (2013 Melbourne)
BRKSEC-2003 – IPv6 Security Threats and Mitigations (2013 Melbourne)
COCRST-2464 – Inside Cisco IT: Making The Leap To IPv6 (2013 Melbourne)
As you can see. IPv6 is a pretty big deal these days at Cisco Live. Then you also have
books, configuration guides etc but this should give you a good start to see what challenges
and considerations you should have when deploying IPv6.
We’re turning up an IPv6 core now. Most of the routing is in place, most of the transit links and loopbacks have been IP’d, and we’re getting ready to deploy IPv6 services to customers.
One of the networks we acquired runs statically-assigned, pulic IPs in /127 subnets. I don’t particularly consider this to be a “good” idea, although it can definitely ease troubleshooting. I much prefer the link-local, auto-assigned approach. It’s easier to deploy and has the added benefit of not being routable. Some of the arguments to run public IPv6 addresses on transit links seem silly. I don’t really need DNS on transit links, and I don’t care if traceroute returns loopback IPs. That just means I log into the router to look at stuff, which I probably will have to do anyway.
We’re doing all of this native. It’s made easier by using link-local automatic addresses for the transit links. We assign loopback IPv6 addresses to all of our equipment, turn inet6 on the transit interfaces, drop the interfaces in the IGP, and we’re done.
Hi!
We get IPv6 natively from both NORDUnet and FUNET, don’t have a major network to look after myself though maybe we start to use BGP in the near future.
Have you gotten the chance to play with IPv6 and BGP?
I really liked Hurricane Electric’s ipv6 certification (http://ipv6.he.net) – also the t-shirt you can get when you’re at the highest level is actually quite nice looking 🙂
Haven’t done that much BGP with v6. I’ve heard of people having issues with the next-hop due to link local addresses. Maybe I’ll do some labs. I’ll have to check out HE 🙂