[Cake] Van Jacobson at netdevconf earlier this week

Dave Taht dave.taht at gmail.com
Mon Jul 16 11:09:38 EDT 2018

On Mon, Jul 16, 2018 at 5:04 AM Jamal Hadi Salim <jhs at mojatatu.com> wrote:
> We are an engineering conference(lanman appears to be academic)
> tied to Linux networking only. To get a sense, last schedule:
> https://www.netdevconf.org/0x12/schedule.html

The MIT folk presenting a trace driven (ugh!) study showing how their
sprout solution worked better than a misconfigured sfq_codel at 150+ms
endemic delay within the celluar network struck me as rather academic.
'course, if that's the 5G network that is going to be inflicted on us,
despite a 1ms underlying mac, then perhaps all the associated
complexity they propose is needed, but I sure as hell hope a design
goal for 5G is less than 20ms of queuing delay in the first place, and
typical physical rtts in that range also.

the MIT folk been beating that trace driven thing to death for years
now (goes back to keith winstein's work years ago), and complaining
about how hard it is to do real work in celluar, and eating up
valuable conference time with something that... well... if you look at
their paper, they *start* with assuming 100ms delays and obscure their
arguments from there.

(mahimahi is cool though)

> For selfish reason I hope we stay small because it is a lot more
> intimate (but i worry because the attendance is growing).
> Mainstream media doesnt fit well with our crowd. We talked to lwn
> but unfortunately they couldnt send anyone.

LF subsidizes lwn's travels. So do other conferences. I would like it
if netdevconf had broader impact (and I wish I was there for a
multitude of presos!)... so I'm awaiting the videos and papers

> They will, however,
> interview Van.


> An attendee wrote up something on Van's keynote:
> https://jvns.ca/blog/2018/07/12/netdev-day-2--moving-away-from--as-fast-as-possible/

That was how I heard of it. A bit lightweight, that.

> Van's keynote I believe is _very influential_. NIC vendors were
> present and I am sure you'll start seeing NICs supporting his ideas
> in the very near future.

I have the general outline already, and I like it. :) An actual API
and some hardware to play with would be great.

> We are working on cleaning/cutting/editing videos - they should
> be up soon. Likely the slides will go up first; then papers before
> you see videos.
> Next conference is in Prague Mar 20-22 2019. We do not accept
> recycled talks - so if you are going on a "tour" start with us.
> You can look at an option for submitting a tutorial perhaps.

sch_cake will be in wide distribution then, but perhaps we'll come up
with something out of it's design worth presenting.

> cheers,
> jamal
> On 15/07/18 11:35 AM, Dave Taht wrote:
> > On Fri, Jul 13, 2018 at 6:57 PM Jonathan Morton <chromatix99 at gmail.com> wrote:
> >>
> >>> On 14 Jul, 2018, at 1:22 am, Dave Taht <dave.taht at gmail.com> wrote:
> >>>
> >>> I look forward to the press coverage and videos of these sessions.
> >>
> >> Did you get anything like that from LANMAN?
> >
> > Lanman is a very small conference. I viewed my preso there as an alpha
> > test for future conferences. At one level
> > going on the road again at this point, to talk to the wisp, 5g, rdk,
> > netdevconf, and ietf folk makes sense, at another, well... lacking
> > travel budget, graphic sense, and patience, makes it intimidating.
> > Still, I am trying to summon up the gumption to go on the road
> > again... or at least do talks in my local area.
> >
> > netdevconf is also a very small conference. Although I think highly of
> > what they are doing (being far more technical and linux grassroots
> > oriented, than LF, and going back to back with ietf in the hope that
> > some folk and ideas bleed across), I was being sarcastic about even VJ
> > getting any coverage from the press. I doubt anyone we know in the
> > tech press was there, not ars, not wired - not even sure if lwn sent
> > anyone.
> >
> > do hope various talks were filmed though.
> >
> >>   - Jonathan Morton
> >>
> >
> >


Dave Täht
CEO, TekLibre, LLC
Tel: 1-669-226-2619

More information about the Cake mailing list