On Thu, Jun 22, 2023 at 1:02 AM Brian E Carpenter <brian.e.carpenter@xxxxxxxxx> wrote:
> 5: the time from consensus is declared until it is sent to the IESG?
Considering that our process makes this a one-person bottleneck for
a person who is supposedly a part-time volunteer with a day job
too, and considering that we have all (I hope) studied queueing theory
for single-server systems, I'm very unsympathetic to complaints
about this compared to all the other steps in the process.
The input within the queue of AD_review is known, but the service time (or waiting time) is unknown, the AD should help make it known to waiting_worker.
If you want to reduce the mean response time of this queue, the best
way would be to send it less work.
The best way can be using Scheduling Techniques, because not all work/draft are having similar load/dependencies/complexity.
As long as the IETF has ~130 active
working groups and a manageable size of IESG, i.e. about 10 WGs per AD,
this isn't going to happen.
So each WG already is scheduling with milestones to send one ready_WGLC_draft each two months or less, so each AD can complete 5 reviews per IETF_Area each weak (i.e. makes about reasonable 20 reviews per month). ADs should inform authors/WG of their reasonable response time/schedule of ietf_area_service.
best wishes,
AB
