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

Re: Proposed IPv6 DNS Discovery Requirements

At Tue, 30 Apr 2002 14:47:20 -0700, Bob Hinden wrote:
> The first set of sections (1-8) does not specify the use anycast or 
> multicast, only unicast using three well know addresses.

Yes, but....

>From the client's point of view, the practical difference between a
well-known unicast address and an anycast address is almost
non-existant, other than removal of the can't use as source address
restriction that folks are already talking about elsewhere in this
thread.  Using well-known unicast addresses still moves knowledge of
the server location into the routing system.  Using well-known anycast
addresses still leaves the client unable to take any useful recovery
action if the server location mechanism has returned answers that are
not useful.

Basicly, take all my ranting about service discovery via anycast
address over the last few days, apply s/anycast/well-known unicast/g
and almost all of it will still parse.  That being the case, I assume
that you're no more interested in reading all that again than I am in
sending it all again, so I won't.

I am still fairly skeptical about the existance of a real need for the
service provided by level 1 compliance, given its limitations (some of
which I've commented on, some of which are commented on in the draft).

Finally, I would urge folks to read RFC 1535 then go back and read
section 3.2 of the draft.  Inferring DNS search paths from site names
is dangerous.  It's not too bad if one takes reasonable precautions,
but the draft would have to spell out those precautions (see the RFC).

> Hope you enjoyed your coffee.

The first cup of coffee recapitulates phylogeny.

IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to majordomo@sunroof.eng.sun.com