On Mon, Jul 15, 2013 at 12:50 PM, Joe Touch <[email protected]> wrote: > FWIW, I don't think transport == congestion control. >
Joe is exactly correct about this. > > There are a lot of other issues in connection semantics, service IDs, etc. > that are integral to the transport layer which are equally critical. > Another example, a large part of fq_codel's gains are over the sort of flow queuing it does, rather than congestion control. Our experience is congestion control is necessary, but not sufficient to deal with bufferbloat (see my recent blog article at: http://gettys.wordpress.com/2013/07/10/low-latency-requires-smart-queuing-traditional-aqm-is-not-enough/ ). > > And I don't think a transport AD needs to be an expert on this, but they > cannot rely on TSVDIR to point out key issues or concerns either; TSVDIR > isn't on every IESG telechat, nor attending every IESG meeting at the IETF. > > TSVDIR aren't "guide dogs" for the (transport-)blind. ;-) > > We can help dig into details, but if the AD doesn't know to put the brakes > on or raise an issue, by the time we get a doc it will be far too late. I don't think the IETF can find people who know everything in depth (the person doesn't exist, probably); but nodding familiarity with a wide range of the issues is desirable. - Jim > > Joe > >
