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

Re: Proposed IPv6 DNS Discovery Requirements

    Date:        Tue, 23 Apr 2002 08:28:43 -0400
    From:        Ralph Droms <rdroms@cisco.com>
    Message-ID:  <>

  | Here's my cut at a list of requirements for DNS Configuration.

I quite like this list, except perhaps ...

  | 10. Hosts have one way to perform DNS configuration

Unless that means only what the notes say ...

  | 10. Hosts shouldn't need to try several different DNS Configuration
  |      protocols to find which one is available on a link

That's fine - a host should only need to try one way, but there should
me multiple possible ways to obtain the config, just as we have for
every other config option for IPv6.

Obviously, the admin of a host must be able to manually config the DNS
servers that are to be used (switch off any autoconfig that the host would
normally do).

And if autoconfig is enabled, we need both the "managed" variety, where
someone actually does configure and run a server that supplies exactly the
info that that administrator says should be supplied (this basically means
discovery via DHCP I think) - but we must not require that, we also need
the option where the host simply figures out what exists, for the cases
where there is no network admin, or the admin simply doesn't care.

  | 11. Hosts and DNS servers can exchange datagrams using site-local
  |      addresses

I'm not sure how that relates to point 11...

  | 11. DNS configuration operates when host and DNS servers are in same
  |      site or in different sites

I wouldn't be including anything about address scoping in a requirements
list, not even in notes, that's a potential part of the solution, not of
the requirements.


ps: while a solution that requires no additional implementation or 
specification would be nice, I certainly wouldn't rule out anything just
on that basis - we have (or should have) the manual option now, and
should have dhcp as an option sometime soon, if not now, the idea for
the third form (configureless autodiscovery) should be that it work well,
and be rationally defined, not just the best we can grab out of the
protocols that exist today.

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