[Rpm] [Bloat] Airbnb
Matt Mathis
mattmathis at google.com
Mon Aug 16 13:00:17 EDT 2021
I think you misunderstood: The metric that I am testing can be computed in
real time within the current NDT (the only wire change is to carry an
additional result in the progress reports) and across our entire archive,
including both web100 and tcp_info data. Furthermore it can be
validated/calibrated by counting actual network rounds, reconstructed from
packet captures. However it does not include the OS contribution, which
can have a huge impact on the responsiveness seen by the user.
The MIT paper boosts another item that has been on my forever todo list:
add automated diagnostic analysis to NDT. Ideally, NDT could localize the
bottleneck:
- Client (e.g. insufficient CPU or buffers)
- Last yard (WiFi)
- Last Mile (Access provider)
- Middle mile (Interconnects)
- Server load
(The original NDT did a limited version of something like this. Most of
the above have well known signatures, except there are some really
weird corner cases.) We could then better inform the user ("Fix your
WiFi"), and label the data in the archive according to the nature of the
likely bottleneck.
Thanks,
--MM--
The best way to predict the future is to create it. - Alan Kay
We must not tolerate intolerance;
however our response must be carefully measured:
too strong would be hypocritical and risks spiraling out of
control;
too weak risks being mistaken for tacit approval.
On Fri, Aug 13, 2021 at 11:59 AM Livingood, Jason <
Jason_Livingood at comcast.com> wrote:
> Great news, Matt! If I may make a recommendation, IMO a standalone working
> latency test would have greater utility than integrating such a measurement
> into an existing test & so it may have the opportunity of more widespread
> usage.
>
>
>
> JL
>
>
>
> *From: *Bloat <bloat-bounces at lists.bufferbloat.net> on behalf of Matt
> Mathis via Bloat <bloat at lists.bufferbloat.net>
> *Reply-To: *Matt Mathis <mattmathis at google.com>
> *Date: *Tuesday, August 10, 2021 at 00:52
> *To: *"dave.collier-brown at indexexchange.com" <
> dave.collier-brown at indexexchange.com>
> *Cc: *bloat <bloat at lists.bufferbloat.net>
> *Subject: *Re: [Bloat] [Rpm] Airbnb
>
>
>
> I am part of the M-Lab team.
>
>
>
> For years we published a "jitter" metric that I considered to be bogus,
> basically max_rtt - min_rtt, (which were builtin Web100 instruments).
>
>
>
> In 2019, we transitioned from web100 to "standard" linux tcp_info, which
> does not capture max_rtt. Since the web100 jitter was viewed as bogus, we
> did not attempt to reconstruct it, although we could have. Designing and
> implementing a new latency metric was on my todo list from the beginning of
> that transition, but chronically preempted by more pressing problems.
>
>
>
> It finally made it to the top of my queue which is why I am suddenly not
> lurking here and the new rpm list. I was very happy to see the Apple
> responsiveness metric, and realized that M-Lab can implement a TCP
> version of it, that can be computed both in real time on future tests and
> retroactively over archived tests collected over the last 12 years.
>
>
>
> This quick paper might be of interest: *Preliminary Longitudinal Study of
> Internet Responsiveness*
> <https://urldefense.com/v3/__https:/docs.google.com/document/d/1HkeJTgwLdbbdHe0dF9SWrpr8Vt7ngnovLpdDTScuUtk/edit*heading=h.sy6m01i6oa9a__;Iw!!CQl3mcHX2A!WTavRmT2GnSUSVpvPCPdmBuNtOJI_zDP1WOHE-jy8tMiXBE-CxDospZUjNote1PXn0DhWQ$>
>
> .
>
> Thanks,
>
> --MM--
> The best way to predict the future is to create it. - Alan Kay
>
> We must not tolerate intolerance;
>
> however our response must be carefully measured:
>
> too strong would be hypocritical and risks spiraling out of
> control;
>
> too weak risks being mistaken for tacit approval.
>
>
>
>
>
> On Mon, Aug 9, 2021 at 5:27 PM Dave Collier-Brown <
> dave.collier-brown at indexexchange.com> wrote:
>
> Knowing programmers, I'd more likely assume latency when it's easiest to
> measure. At a guess, either
>
> - average of upload or download
> - average of download (as it comes first)
>
> --dave (call me "Wally" (:-)) c-b
>
>
>
> On 2021-08-09 4:24 p.m., Jonathan Morton wrote:
>
> On 9 Aug, 2021, at 10:25 pm, Dave Collier-Brown <dave.collier-brown at indexexchange.com> <dave.collier-brown at indexexchange.com> wrote:
>
>
>
> My run of it reported latency, but without any qualifiers...
>
> One would reasonable assume that's idle latency, then.
>
>
>
> - Jonathan Morton
>
> _______________________________________________
>
> Bloat mailing list
>
> Bloat at lists.bufferbloat.net
>
> https://lists.bufferbloat.net/listinfo/bloat <https://urldefense.com/v3/__https:/lists.bufferbloat.net/listinfo/bloat__;!!CQl3mcHX2A!WTavRmT2GnSUSVpvPCPdmBuNtOJI_zDP1WOHE-jy8tMiXBE-CxDospZUjNote1Oa4DQV1Q$>
>
> --
>
> David Collier-Brown, | Always do right. This will gratify
>
> System Programmer and Author | some people and astonish the rest
>
> dave.collier-brown at indexexchange.com | -- Mark Twain
>
>
>
> *CONFIDENTIALITY NOTICE AND DISCLAIMER** : This telecommunication,
> including any and all attachments, contains confidential information
> intended only for the person(s) to whom it is addressed. Any dissemination,
> distribution, copying or disclosure is strictly prohibited and is not a
> waiver of confidentiality. If you have received this telecommunication in
> error, please notify the sender immediately by return electronic mail and
> delete the message from your inbox and deleted items folders. This
> telecommunication does not constitute an express or implied agreement to
> conduct transactions by electronic means, nor does it constitute a contract
> offer, a contract amendment or an acceptance of a contract offer. Contract
> terms contained in this telecommunication are subject to legal review and
> the completion of formal documentation and are not binding until same is
> confirmed in writing and has been signed by an authorized signatory.*
>
> _______________________________________________
> Bloat mailing list
> Bloat at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat
> <https://urldefense.com/v3/__https:/lists.bufferbloat.net/listinfo/bloat__;!!CQl3mcHX2A!WTavRmT2GnSUSVpvPCPdmBuNtOJI_zDP1WOHE-jy8tMiXBE-CxDospZUjNote1Oa4DQV1Q$>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/rpm/attachments/20210816/f1c04d51/attachment.html>
More information about the Rpm
mailing list