Re: How IPv6 host gets DNS address

[Ralph Droms <rdroms@cisco.com>]

Ohta-san,

Your use of English was almost too subtle for me.  I had to read your
problem statement a couple of times before I realized you were turning
around my basic assumption.  So, rather than assuming stateless address
autoconfiguration and trying to solve the PTR record population problem, we
should solve the PTR problem by picking the right address configuration
process.

I believe there are some sites who have come to the conclusion that DHCPv6
address assignment is the right deployment model, exactly because it avoids
the PTR record problem as well as giving network administrators reliable
information about mappings between devices and IPv6 addresses.

- Ralph

At 10:28 AM 11/6/2003 +0900, masataka ohta wrote:
>Ralph;
>
>>We have other significant problems to solve to complete the IPv6
>>name-and-address management story - most importantly at this point, how do
>>we arrange for population of PTR records by roaming hosts using stateless
>>address autoconfiguration?  We should be spending our cycles on those
>>problems...
>
>Wrong problem.
>
>The problem to be solved (if not yet solved) is
>
>         how do we arrange for population of PTR records by roaming
>         hosts using address autoconfiguration
>
>and DHCP, rather than DHCP-lite, is, it seems to me, the only
>way to go.
>
>>I strongly believe, based on
>>implementation experience, that the implementation complexity of DHCPv6
>>for host configuration is *not* prohibitive.
>
>Compared to the implementation complexity of ND, yes.
>
>The easiest, simplest and fastest way to go is to remove ND,
>a full set of useless features, entirely.
>
>                                                 Masataka Ohta
>
>
>#----------------------------------------------------------------------
># To unsubscribe, send a message to <dnsop-request@cafax.se>.

#----------------------------------------------------------------------
# To unsubscribe, send a message to <dnsop-request@cafax.se>.