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

[idn] RENAME function for any ACE implementation.




> one of the interesting things about this release is that NSI-Registrar is
> apparently not an approved participant in the testbed run by
NSI-Registry....
>
Interesting observation...

> There seems to be a great rush to get the "ownership" of customers, on
> whose behalf names can be registered later.
> But our task is to make sure there is ONE standard by which to register
> them in the real DNS.
> And soon.

I think this is right and we should refocus our discussions towards
establishing the ultimate form, transition form and implementation of
multilingual names into the DN system.  There will be people trying to get
the market for the "registration" part, and we can do nothing here to stop
that.  But we are in a position to steer how the names will actually be
resolved.

This brings me back to some of my proposals to converge the two main streams
of implementation (as I see it) 1) bit-flag identifier (DNSII, IDNE, UDNS)
and 2) ACE format.  A hybrid approach will see that an ACE format be used as
the immediate transition so that lower level DNSs do not need to change,
with the bit-flag identifier being the ultimate transport protocol as it
would be much more efficient (no conversion is required anywhere).  The
problem with any ACE implementation however is that client side applications
might not be ready (client side must be able to convert to ACE before
sending to wire), therefore responses to these requests will be
inconsistent.  During the RACE and BRACE discussions however, there were no
suggestions of how we could deal with this issue.

Therefore, I have proposed a mechanism called RENAME (inside the TRACE
draft) which automatically CNAME an incoming non-ACE multilingual request
into its ACE format before doing a resolution.  This resolves the issue with
the client end without too much of a hassle for the server.  And CNAME is a
will understood RR by any DNS.

In fact we (at Neteka) have tried to implement this by patching a BIND and
it works perfectly well with any kind of ACE.

Edmon

>
> --
> Harald Tveit Alvestrand, alvestrand@cisco.com
> +47 41 44 29 94
> Personal email: Harald@Alvestrand.no
>