I agree with Lixia's comment: there is still important work to be done in the analysis of routing table growth. (Very good work has already been done in this area, but some existing hand-waving needs to be converted to more rigorous models.)
I don't have an opinion of whether this is done in GROW or elsewhere, but it needs to be done with this community--SP's, vendors, academics, working together. Cheers, -Benson > -----Original Message----- > From: David Kessens [mailto:[EMAIL PROTECTED] > Sent: Tuesday, March 20, 2007 4:16 PM > To: Lixia Zhang > Cc: Ron Bonica; [email protected]; Dan Romascanu > Subject: Re: [GROW] Proposal to close grow working group > > > Lixia, Henk, > > On Tue, Mar 20, 2007 at 06:54:22AM -0700, Lixia Zhang wrote: > > > > Copying Henk's word, > > We have no problem in offering a path forward for the mrt draft, > including timelines - either as individual submission or through a Ops > area working group (although that is a less likely route considering > the desire for quick publication) > > > I would like to make it an even bit stronger > > (I've sent a similar msg to the ADs earlier): I see values > of keeping > > GROW for another while. In addition to the need of finishing MRT > > work, this group seems appropriate place to help with the coming > > effort (which is supposedly topics of Routing area meeting on > > Thursday) on the need to better understand BGP table growth and on > > what can be done to improve BGP dynamics/churn, topics that > have been > > covered by GROW in the past. > > > > taking BGP update dynamics as example, well defined protocol > > mechanisms exist to dampen routing churns (MRAI, flap > damping), but > > dampening is probably seeing reduced deployment, and even > MRAI is not > > universally turned on (I showed some data at IETF61 GROW meeting, > > http://www3.ietf.org/proceedings/04nov/index.html). So it > seems to > > me these issues are not so much of being on IDR plate but > on GROW's. > > We need a clear understanding of deployment status of existing > > protocol mechanisms and the reasons/causes of why some > places don't > > do it. There can be need for new protocol knobs (IDR job), but it > > seems to me a pre-requisite is a good understanding of issues with > > existing knobs. > > My concern is that I see a working group that has finished it's > charter and where energy levels to do the work as opposed to doing the > easy thing of proposing new work are quite low. > > Note that I believe it will be useful to take elements from the grow > charter to a proposed OPs working group. We still need a place in IETF > where we can discuss these kind of issues. > > David Kessens > --- > > _______________________________________________ > GROW mailing list > [email protected] > https://www1.ietf.org/mailman/listinfo/grow > _______________________________________________ GROW mailing list [email protected] https://www1.ietf.org/mailman/listinfo/grow
