Martin Stiemerling <[email protected]> wrote: > > The IESG makes a write-up about the desired expertise of Area Directors > in general and also for each area. This write-up is sent to the NOMCOM. > > There is now a new write-up for the upcoming deliberations of the > NOMCOM, taking the comments received in the IETF-86 area meeting into > account: > http://trac.tools.ietf.org/group/iesg/trac/wiki/TransportExpertise >... > Please send your feedback either to the TSV-Area list, directly to > Spencer and/or me, or to the IESG until > > ***Thursday, 2013-07-18 10am EDT***
The IESG wants to finalize their input to NOMCOM on July 18. There will be an additional period for comments _after_ it goes to NOMCOM... But this is worth discussing on this list now. For at least the last four years, NOMCOM has had serious trouble finding candidates that meet the desired expertise. For now I'll concentrate on this paragraph: ] ] A Transport AD should have a broad understanding of core end-to-end ] transport topics such as congestion control, flow control, real-time ] transport protocols, NATs and firewalls, and related topics such as ] storage protocols. It is not necessarily important to be an expert ] in any of these, as it is much more important to be knowledgable in ] the principles of these transport topics, such as congestion control ] and congestoin management, Intserv and Diffserv. A Transport AD ] should have good relationships with the topic experts in the ] Transport area and also other areas, which in turn requires good ] soft skills. This is already an improvement over last year's version, which listed a number of specifics which I, at least, had suggested were no longer required or of limited duration. But I direct everyone's attention to David Harrington's message about the problems he faced as Transport AD: http://www.ietf.org/mail-archive/web/ietf/current/msg77860.html The whole post is well worth reading, but I particularly note: ] ] You'll need enough understanding of the TSV issues to be able to ] spot bad transport-related decisions in documents coming from ] elsewhere in the IETF. You (and your co-AD) are effectively the ] reviewer of last resort for TSV issues. If you don't understand ] control loops, congestion control techniques, etc., you will ] NEED to rely on your directorate for assistance in this role. I suggest that "understand control loops" deserves specific mention in the desired expertise. Personally, I'd add "hysteresis in control loops". "Congestion control techniques" is definitely something the NOMCOM should look for candidates to understand. But to me, that wording is open-ended; and we don't need an expert on all such techniques. I think we need someone who understands the basic "additive increase; multiplicative decrease" mechanism of vanilla TCP; but that other _control_ techniques are too varied -- a Transport AD will need to have Directorate members help with those. I'd like to see some mention of understanding "congestion signaling". We've evolved quite a bit from the days when packet drop was the only congestion signal. ECN is in active, if limited, use. There are a number of delay-based indicators now: mostly inferring queueing delay from transit time, but recently actually measuring time-in-queue (such as CoDel and PIE). All of these signals involve trade-offs, if not outright guessing; so we shouldn't look for expert-level understanding, but a basic understanding of congestion signals seems important to me. David Harrington dwells at length on his experience with the Transport Directorate. We might tweak the last sentence of the paragraph I quoted from the desired expertise: " " A Transport AD should have good relationships with the topic " experts in the Transport area and also other areas, which in turn " requires good soft skills. to indicate actual management skills in keeping the Directorate functioning well. As for the other items listed in the desired expertise: " " topics such as congestion control, flow control, real-time " transport protocols, NATs and firewalls, and related topics " such as storage protocols. I'd really prefer that others comment first on what level of understanding we should ask for. They are all interesting (though I personally regard NATs as damage to route around), but I don't know how essential they are to a Transport AD. Recall that we _need_ to make the NOMCOM's job possible! -- John Leslie <[email protected]>
