12 problems: 1 Too Slow
Erik.Nordmark at sun.com
Thu Jan 16 19:27:19 CET 2003
> ISSUE 1: Too Slow
> We hear assertions from 3rd parties that the IETF is too slow
> (ironically, the same complaint we used to make about ITU and ISO).
> This is cited as a major reason for doing work in a closed
> vendor consortium.
> The fact is that today it typically takes 18 months for a
> document to progress from a -00 draft to a Proposed Standard RFC,
> up from 6 months in 1993.
> In that period the IETF has tripled in active participants, more
> than doubled the number of WGs, and its demography has changed
> dramatically. Its ouput (measured in RFCs) has at least tripled.
> Some slowing down seems inevitable, but a factor three is slightly
It would be interesting to have data on the distribution of
the time from idea to RFC. How much time does the BoF/chartering stage take?
How much time in the WG before the draft is first sent to the IESG?
How much time in IESG and RFC-editor?
That might help root-cause the slowness for the different components of the
I think the list has talked a fair amount about the tail end delays (IESG).
Margaret brought up the issue of deciding between two choices which might
be a potential root cause for the WG taking a long time.
Looking at even well-managed WGs doing a selection process this doesn't
surprise me (develop requirements doc, morph that into a selection
criteria, issue "request for protocols", evaluate against selection criteria,
make the choice).
Are there more WGs doing such selections (even if not as thurough as
the above list) that there were say 5 years ago?
The fact that we have more technology to build upon and reuse might be a
natural reason for there being more selections and less original engineering.
I'm not saying that this is bad, but if it is a significant contributor
to unnecessary delay it would be useful for the IETF to try to understand
how to make selections more efficiently.
Examples of what I'd consider "selections" range from relatively simple things
like WGs choosing between running over TCP, SCTP, or both,
to more complex things like the AAA WG, and instant messaging.
And I'm sure there are lots of other examples of this.
More information about the Problem-statement