RE: getaddrinfo address ordering [Re: IPv6 transition architecturediscussion]

[Pekka Savola <pekkas@netcore.fi>]

On Wed, 27 Nov 2002, Christian Huitema wrote:
> > > It will currently always fall back to v4 if one wrote it correctly:
> > >  - tries AAAA (if dns returned it)
> > >  - tries A (if dns returned it)
> > 
> > Sure.. but in this case www.google.com will not have AAAA records in,
> > well, about 5-10 years.  They have no incentive to do so, _at all_ --
> > quite the contrary -- as their v4-enabled users would get worse
> service.
> 
> A nice point of 6to4 as a transition strategy is that its performance
> are guaranteed to be very close from those of the underlying IPv4. A
> normal evolution for sites like Google would be to multihome and expose
> both a native IPv6 connection and a 6to4 connection. Address selection
> rules will guarantee that "transition" users pick the 6to4 destination,
> while native users pick the native address. This will largely deal with
> your performance concern.

This would help only if all v6 users, v6-native included, would enable
6to4 as an "optimization technique", as proposed previously.

Doesn't seem like a scalable solution.

(why v6-native?  v6 native connection, here meaning non-6to4 global 
addresses, cannot be guaranteed to reach google in an optimal fashion -- 
quite far from it.)

Another transition strategy would be to _not_ use any native v6 _at all_
yet, only 6to4 -- or feed the hosts only native more specific routes to
those networks that are "well-connected" (where you wouldn't use 6to4).

Reminds too much of Jim Flemings ideas, too scary to even contemplate.

-- 
Pekka Savola                 "Tell me of difficulties surmounted,
Netcore Oy                   not those you stumble over and fall"
Systems. Networks. Security.  -- Robert Jordan: A Crown of Swords