[Ecn-sane] Fwd: [bbr-dev] Re: BBRv2 v2alpha branch rebased from 5.4 to 5.10, supports all CCs and E2 GCE VMs

[Ecn-sane] Fwd: [bbr-dev] Re: BBRv2 v2alpha branch rebased from 5.4 to 5.10, supports all CCs and E2 GCE VMs

[Dave Taht]

what's the state of SCE nowadays? Has anyone ported it to bbrv2 yet?

I was going to take a stab at D*....

---------- Forwarded message ---------
From: 'Neal Cardwell' via BBR Development <bbr-dev@googlegroups.com>
Date: Fri, Jan 15, 2021 at 3:29 PM
Subject: [bbr-dev] Re: BBRv2 v2alpha branch rebased from 5.4 to 5.10,
supports all CCs and E2 GCE VMs
To: BBR Development <bbr-dev@googlegroups.com>


Based on some quick feedback from the user community (thanks!) we have
fixed a few minor issues with yesterday's release, and pushed an
update to the v2alpha branch in the github repo. The fixes can be
found at the latest tagged release:
  https://github.com/google/bbr/releases/tag/v2alpha-2021-01-15

The changes since yesterday's release:

(1) a bug fix for the warning that has been reported by a few users:
  WARN_ON_ONCE(bbr->prior_cwnd == 0);

(2) a fix for a compilation error caused for users of the following
  .config option:
    CONFIG_BPF_SYSCALL=y

(3) enabled the .config option CONFIG_BPF_SYSCALL=y to catch further issues
  of this kind

best,
neal

ps: as always, you can find the instructions to download and build
Linux TCP BBRv2 alpha here:
  https://github.com/google/bbr/blob/v2alpha/README.md


On Thu, Jan 14, 2021 at 6:03 PM Neal Cardwell <ncardwell@google.com> wrote:
>
> Hi all,
>
> By popular demand, we have rebased the BBRv2 v2alpha branch from 5.4 to 5.10.
>
> The BBRv2 code in this release should be functionally identical to the previous release, since in this rebase we only made the minimal changes to the BBRv2 and TCP patches necessary to get the code to compile.
>
> There were two changes of note we incorporated during the rebase. The kernel .config generated by the gce-install.sh build/install/boot script now has some added features:
>
> (1) it builds all available congestion control modules, to facilitate comparisons (a request made recently on the bbr-dev list here)
>
> (2) it has UEFI support needed to allow booting on the new E2 "cost-optimized general purpose" VM type on Google Cloud
>
> You can find the list of rebased commits here.
>
> As always, previous versions of the v2alpha release can be found using the git tags in the Google bbr repository.
>
> best,
> neal
>
--
You received this message because you are subscribed to the Google
Groups "BBR Development" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to bbr-dev+unsubscribe@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/bbr-dev/CADVnQym7fxYpvC%3Dh38Ga9ORWsOXcS2Mk3jUecKWtBSc7%2BneEUw%40mail.gmail.com.


-- 
"For a successful technology, reality must take precedence over public
relations, for Mother Nature cannot be fooled" - Richard Feynman

dave@taht.net <Dave Täht> CTO, TekLibre, LLC Tel: 1-831-435-0729

Re: [Ecn-sane] Fwd: [bbr-dev] Re: BBRv2 v2alpha branch rebased from 5.4 to 5.10, supports all CCs and E2 GCE VMs

[Jonathan Morton]

> On 16 Jan, 2021, at 1:51 am, Dave Taht <dave.taht@gmail.com> wrote:
> 
> what's the state of SCE nowadays? Has anyone ported it to bbrv2 yet?

We haven't attempted to do anything with BBR yet, instead trusting that any results using conventional AQM and/or L4S will translate reasonably easily.  Instead we're focusing on conventional window-based algorithms, getting the reference qdiscs right, and pointing out enough flaws in L4S so that it stops being an obstacle to IETF progress.

We *have* got a big vote of confidence in the form of funding from both RIPE and NGI Pointer.  That gives us time and freedom to work on SCE without too many unnecessary distractions.  One of the things we're doing with that is building a decent test harness.

 - Jonathan Morton

Re: [Ecn-sane] Fwd: [bbr-dev] Re: BBRv2 v2alpha branch rebased from 5.4 to 5.10, supports all CCs and E2 GCE VMs

[Dave Taht]

On Fri, Jan 15, 2021 at 5:19 PM Jonathan Morton <chromatix99@gmail.com> wrote:
>
> > On 16 Jan, 2021, at 1:51 am, Dave Taht <dave.taht@gmail.com> wrote:
> >
> > what's the state of SCE nowadays? Has anyone ported it to bbrv2 yet?
>
> We haven't attempted to do anything with BBR yet, instead trusting that any results using conventional AQM and/or L4S will translate reasonably easily.  Instead we're focusing on conventional window-based algorithms, getting the reference qdiscs right, and pointing out enough flaws in L4S so that it stops being an obstacle to IETF progress.

It would be really nice to compared bbrv2 instances. But I've been
saying that since the beginning.

> We *have* got a big vote of confidence in the form of funding from both RIPE and NGI Pointer.  That gives us time and freedom to work on SCE without too many unnecessary distractions.  One of the things we're doing with that is building a decent test harness.

Very happy to see that.

The thing is, so far as I know, the basic brick-wall aspect of the aqm
hasn't changed much, especially when FQ is involved, have the aqm
implemantations changed at all for SCE?

>  - Jonathan Morton



-- 
"For a successful technology, reality must take precedence over public
relations, for Mother Nature cannot be fooled" - Richard Feynman

dave@taht.net <Dave Täht> CTO, TekLibre, LLC Tel: 1-831-435-0729