Re: DNS RRTYPEs, the difficulty with

On 2/27/2012 11:57 AM, Murray S. Kucherawy wrote:

To accommodate user interface limitations, the RFC that defined SPF (2006) 
included use of TXT as a "backup" to enable a period of transition.  
Collected evidence shows that a lot of clients do query the type, but fewer 
than 5% of participating sites ever publish it even after six years.

In an ideal world I'd also like to see us move in the direction that RFC5507 
promotes, but it seems we still aren't there yet.


It's important to understand the history here. The problem of "old
software chokes on new RR types" was a known issue way back when, and
3597 came out in Sept. 2003 to help address it. When 4408 came out
defining SPF in April of 2006 support for 3597 was still being rolled
out, and was nowhere near the point where it could be safely assumed in
even a meaningful minority of the deployed base.

The error for SPF came with the TXT bootstrap version. It's easy to see
now (whether it was easy to see then or not is water under the bridge at
this point) that this was a mistake. As painful as it would have been
back in 2006 to wait for support for 3597 and/or the SPF RRtype to
propagate, we would be in a much better situation *now* if we had waited.

So, what we need to do is learn from that experience. 8.5 years later
support for 3597 is a very reasonable thing to expect, and with AAAA,
DNSSEC, etc. we're well past the era where hidebound DNS software is an
acceptable operational model.

Do I believe that insisting on new RRtypes will be a pain-free
experience? Of course not. But if we don't insist on it at this point,
every day we delay it becomes more and more painful, not less.

To circle back to my suggestion that SRV be an integral part of any new
HTTP effort, it's 12 years old, so predates both of the above. :)


Doug

-- 

        It's always a long day; 86400 doesn't fit into a short.

        Breadth of IT experience, and depth of knowledge in the DNS.
        Yours for the right price.  :)  http://SupersetSolutions.com/

_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf