Date:        Sat, 09 Jun 2001 08:57:28 -0700
    From:        Randy Bush <[EMAIL PROTECTED]>
    Message-ID:  <[EMAIL PROTECTED]>

  | e.g. in the absense of rapid renumbering and gse or other non-v4 routing,
  | what need is sufficiently important to justify a6?

Unless something happened to routing I'm not aware of (which is
certainly possible), we still need rapid renumbering.

So, it isn't in the absence of that...

  | if there is none
  | currently, but one arises in the future (and i sure hope something does
  | arise for at least routing), then, if dns mechanisms are needed, appropriate
  | ones can be specified.  and those could be a6, i can't prejudge.

You're proposing that once we get millions of real operational nets
running v6, that we'll be able to go change the format of the DNS record
that is used to find them all?   What mechanism would you suggest exists
that makes that feasible?

  | in the meantime, it would be a real bummer if the current a6 spec, for which
  | there seems to be little documented actual need, was to prejudice designs
  | for critical problems such as routing because some interesting approach
  | would not work with a6.

It truly would indeed.   Especially as anything that works with AAAA works
with A6, given that A6 0 and AAAA are isomorphic.   So you're suggesting
the possibility of something which wouldn't work with any proposed DNS
lookup mechanism - that would be a pain.

But to me this looks like FUD?   Where's the documented actual design that
is being prejudiced by A6?

The reason for switching to A6 now, even if we tell people to only ever
put A6 0 in their zone files, is because it makes it possible to use it
later.  If it turns out to be unecessary, then we end up carrying around
one extra byte with each address in DNS packets (a 6% overhead on the
addresses themselves, less than that once all the rest of the DNS overheads
are added (TTL, class, record type, ...) are included in the RDATA).

kre

--------------------------------------------------------------------
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 [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to