Re: [sip-overload] [dispatch] Chartering SIP Overload

Just following up here...

I sent a private e-mail to Robert, Lars and Magnus that might better have 
gone to the mailing list ...

My point was that

- I understand why TSV detours congestion avoidance drafts through 
Experimental, because we already have congestion avoidance mechanisms that 
actually work, so we should be cautious about dorking with them, but

- our standards-track SIP overload mechanisms have already been observed to 
fail on multiple live networks (I've heard this from two tier-one 
operators), so

- it seems appropriate to use a less restrictive strategy here - we 
shouldn't encourage people to ignore our maturity levels completely, and we 
shouldn't use our maturity levels to encourage people to continue deploying 
standards-track mechanisms that have already been observed to fail, and are 
not self-correcting when they fail, on production networks.

That fails the "first, do no harm" test. IMO, of course.

Spencer, whose first working group was a TSV-classic working group (PILC 
with Aaron Falk), and whose first RFC credit was on TCP congestion avoidance 
behavior ("TCP over Satellite", edited by Mark Allman)

>
> On Jan 18, 2010, at 8:04 AM, DRAGE, Keith (Keith) wrote:
>
>> I am worried that we are raising the bar on standards track documents.
>>
>> I am certainly aware that some of the mechanisms described in the 
>> documents have been prototyped and have been shown to work. With  that, 
>> the documentation merits standards track deliverables.
>
> Judging from the number of "We have implemented this" responses, it  might 
> almost be ready to go from Proposed to Draft Standard.
>
> --
> Dean
> _______________________________________________
> dispatch mailing list
> dispatch@ietf.org
> https://www.ietf.org/mailman/listinfo/dispatch 