Re: AXFR over UDP is available

[Edward Lewis <[e-mail]>]

At 9:10 +0900 2/9/08, Masataka Ohta wrote:
>Edward Lewis wrote:
>
>>  http://www.ietf.org/internet-drafts/draft-lewis-axfr-over-udp-00.txt
>
>I'm afraid you completely misunderstand the previous discussion.

No, but I clearly was judging the discussion through different eyes.

>First of all, you failed to convince us that changes to enable
>AXFR over UDP is desirable. But, let's ignore it for the rest of
>my message.

I wasn't trying to.  I wanted to get flesh out the protocol to see 
first if it is viable.  It's an engineering document, not a marketing 
one.

The initial discussion convinced me that it is worth proposing.  If 
the extension is viable, then a use case can be added.

The reason I went forward on "weak evidence" of demand is that I know 
that if there's AXFR over UDP in a separate document, it will get the 
traction of people  issuing RFPs for DNS and DNS-like service.  There 
is a tendency for folks to do a quick search of titles for 
capability, rarely/never have I seen any one go inside an RFC and 
profile a set of services.

>RFC1995 says:
>
>    When an IXFR request with an older version number is received, the
>    IXFR server needs to send only the differences required to make that
>    version current.  Alternatively, the server may choose to transfer
>    the entire zone just as in a normal full zone transfer.

The problem with the above is that it either means:
1) You have to go to TCP
2) You attempt to do the transfer over UDP - but fall into the 
problems of mis-ordered packets, etc. (which I disucss in one of the 
two drafts, I forget which already)

We'd have to clear that up in IXFR.

>As full zone (AXFR) over UDP is already available with IXFR, the only
>thing we need to support AXFR over UDP is to disable differences (IXFR)
>over UDP.
>
>That is, an configuration option on IXFR servers to disable UDP
>differential transfer and to encourage UDP full transfer is just
>fine.
>
>An IXFR server which is incapable of any differential transfer is
>also fine.
>
>IXFR clients, against zone administrators policy to allow differential
>transfer, insisting on full transfer can ignore UDP differential
>transfer responses and initiate TCP AXFR, which causes no extra packet
>exchanges.
>
>There is absolutely no protocol work left.

Except for what you prescribed in your message and clear up my note 
about the RFC 1995 passage.

-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                                [phone]
NeuStar

Mail archives, backups.  Sometimes I think the true beneficiaries of
standards work are the suppliers of disk drives.

--
to unsubscribe send a message to [e-mail] with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/namedroppers/>