[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Mutli6 meeting in Vienna

On vrijdag, mei 16, 2003, at 13:10 Europe/Amsterdam, Brian E Carpenter wrote:

Supporting single homed correspondents is essential but I don't care
much for single homed hosts in a multihomed site. Upgrade already. It's
not like there is a huge installed base of stuff that isn't supported
any more in IPv6.


Well, pretty much every operating system image shipped today includes
RFC 2460 support. So I suspect the installed (but not enabled) base
is in fact huge.




I agree, the operating systems are in good shape. Applications and 
service providers are much more problematic, not to mention the DNS.




At least in a large corporate network, routers and
hosts are *not* upgraded in a coordinated way. So 'upgrade already'
won't fly. We need incremental deployment; that has always been the
guiding principle for IPv6 coexistence.




Well I'm not a huge fan of what has been happening with IPv6 so far. 
Why is it that the IETF can standardize something like A6 DNS records 
that turns out undeployable (big surprise) but at the same time over 
the past 10 years fail to come up with a way for IPv6 hosts to get at a 
DNS server (note: preferably not "discover") without using stateful 
mechanisms? But then if an IPv6 host could query an IPv6 nameserver 
that IPv6 nameserver wouldn't be able to reply without going back to 
IPv4 because the root and TLD infrastructure is oblivious to v6. And 
why exactly? For a large part because we don't want to break hosts that 
can't handle packets that are bigger than half a kilobyte. That's 
right, the amount of data an average computer these days can transmit 
in 50 microseconds.



And why choose either the IPX approach where the MAC address is part of 
the address (simple) or the Appletalk approach where hosts 
automatically find a free address (keep addresses short) when you can 
do both at the same time, adding complexity to large addresses, so 
nobody gets to be happy?



People who like to be conservative when building networks don't get 
what they want from the IETF because there are too many standards and 
*way* too many drafts (and those are often implemented so they can't be 
ignored), but at the same time people who like to be at the leading 
edge don't get what they want either because it takes forever to get 
anything done and everything has to be baby steps.



I'm sure I had a point I wanted to make but it's late so it'll have to 
wait.



Iljitsch