Re: [dnsext] Fwd: New Version Notification for draft-ah-dnsext-rfc1995bis-ixfr-02

[Brian Dickson <[e-mail]>]

On Fri, Jun 17, 2011 at 9:29 AM, Edward Lewis <[e-mail]> wrote:

> If the query type is IXFR, the response will be an IXFR response that
> contains all of the records in one (UDP) datagram, which functionally is the
> same as a AXFR having all of the records appear in multiple DNS messages
> over a stream - the same for only small to very small zones.  The response
> is limited to 512 bytes or whatever is in the EDNS0 buffer size of the
> requestor.

IXFR does not need to be over UDP. In RFC 1995, it even describes falling back
to doing IXFR over TCP, if IXFR over UDP fails for anything other than
a few specific
RCODEs.

If the IXFR query is sent over TCP, the answer will either be an IXFR
(for real), or
an IXFR which includes the whole zone.

IXFR-ONLY is designed specifically to handle the latter case,
preventing the "or..."
from being done.

There are plenty of zones that require IXFR-ONLY in order to
efficiently and effectively
process zone updates.

The motivation and justification for this is clear, IMHO. If there is
no workaround possible,
which I assert is the case for some zones, then IXFR-ONLY is required.

I see no reason not to include it in the protocol specification, if it
is going to be implemented
by more than one implementation. And I assert that the latter is
certainly the case.

Brian