Changing stature of the RFC 'brand'

john.loughney@nokia.com john.loughney@nokia.com
Tue, 10 Dec 2002 09:51:52 +0200


Hi all,

One thing that I have been thinking about, and I'd appreciate comments
from those with longer histories than mine, is that the RFC 'brand' =
seems
to have changed.  It seems that there is a higher bar for drafts to=20
reach nowadays.  It seems that the review process for Proposed Standard =
is
much more strict.  Even Informational RFCs are reviewed much closer =
these
days.  Additionally, protocol extensibility has been tightened quite =
much,
simple IANA registry practices for many protocols now require IESG
action.  The net effect of this is that the IESG work load has =
dramatically
increased.

This has a side-effect that it takes longer for protocols to be =
developed, and=20
we get less deployment information.  I am not sure if this is from =
external=20
forces or internal forces (my hunch is more internal forces). =20

I would be interested to have a discussion on this.  What do we see as =
the
correct target for a Proposed Standard; how much review is enough & how
much is too much?  Do we want to hit 80% of the target or 95% of the =
target
(remember that the last 15% may take an extra year or two).  Do we want
to design protocols that are nearly bullet-proof or do we want to work
on protocols that can be updated once we get more operational =
experience.

I do have a worry that IETF will start making protocols which are long
delayed, not fully deployable and not fully implementable.  In my past,
I worked on implementing the CoreINAP protocol (core set of Intelligent
Network Application Protocol).  It was a beast to try to implement; huge
amount of problems related to specifying something fully before doing
a reality check to see if something is implementable; and no real need =
for
more that half of the protocol features.  This is where I hope we are =
not
going.

So, in summary, maybe we need to check if our 'customers' really need us
to get it 95% right or are they willing to settle for less, especially =
if
it means getting the protocols done quicker.  I agree with Marshall that
in many respects, timely-ness is very important.

John