Re: [v6ops] PI [ULA draft revision #2 Regarding isolated networks]

[Mark Andrews <[e-mail]>]

In message <[e-mail]>, Philip Homburg writes:
> In your letter dated Sat, 31 May 2014 12:41:45 +0200 you wrote:
> >PA with automatic network (re-)numbering and multihoming with multiple
> >PA networks already works today, and will really work much more pleasantly
> >than PI *for those networks* as soon as we've sorted out some of the 
> >remaining kinks (like source-address selection with SA failover).
> >
> >Maybe you should step down from your "I have PI, I like it, everybody must
> >have PI" soapbox and actually look at what, for example, homenet has 
> >achieved in the last years.  This stuff looks complicated (and under the
> >hood, it is), but the end user experience "take this box, plug in a number
> >of ISPs, things work, no further configuration is needed(*)" is nothing you
> >can match with a PI network.
> 
> I can see how you can do do multiple PA prefixes client side. Done that
> for years now. Even with different routers providing the upstreams. No proble
> m
> there.
> 
> But I have nothing to update my DNS zones. How do I reflect which links 
> are up or down? Is there even a draft for that? What's the BCP for TTL
> values, DNSSEC, etc?

You have UPDATE + TSIG or SIG(0).  This is basically what Microsoft
do with Active Directory except they use GSS-TSIG for about 15 years
now.  UPDATE and DNSSEC just work and have for over a decade now.
Happy Eyeballs has proved that you don't need change the DNS for
link state changes.  IPv4 vs IPv6 is no different to PA1 vs PA2.
HE is all about the client working around a dead link.

> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]