Re: [Bloat] this explains speedtest stuff

[Sergey Fedorov <sfedorov@netflix.com>]

[-- Attachment #1: Type: text/plain, Size: 4065 bytes --]

>
> One thing I've been seeing is way too many articles basically talking
> about traffic increases (or not), and how many are videoconferencing from
> home... but no metrics of import to videoconferencing folk. I've been
> engaged in a conversation about increasing a certain videoconferencing
> platform's default jitterbuffer to... wait for it... *1 sec* - based on how
> badly LTE was performing for some people
> so if you control your own videoconferencing tools, collecting more
> metrics there would be usef
> One very interesting metric a netflix streamer could be collecting is
> differences in tcp RTT (assuming you slowed traffic down in europe,
> especially, to a lower quality?), hour by hour, day by day.

All good points and I generally agree with your observation on lack of good
resources/recommendations to improve latency-sensitive network
interactions, especially as those are becoming critical for the users.
I hope we'll be able to share some of the
findings/observations/recommendations based on our experience (white paper,
blog etc), but no hard promises at this point.

SERGEY FEDOROV

Director of Engineering

sfedorov@netflix.com

121 Albright Way | Los Gatos, CA 95032



On Thu, Apr 23, 2020 at 1:27 PM Dave Taht <dave.taht@gmail.com> wrote:

>
>
> On Thu, Apr 23, 2020 at 1:15 PM Sergey Fedorov <sfedorov@netflix.com>
> wrote:
>
>> Hi folks,
>>
>> No need to put more pressure - I've seen Jonathan's suggestion and it
>> makes a lot of sense to add the option to deep-link to an expanded version
>> with all detailed parameters shown.
>> This will be added some time this quarter (Q2), but not within a few next
>> weeks.
>>
>
> that's wonderful, thanks!
>
> One thing I've been seeing is way too many articles basically talking
> about traffic increases (or not), and how many are videoconferencing from
> home... but no metrics of import to videoconferencing folk. I've been
> engaged in a conversation about increasing a certain videoconferencing
> platform's default jitterbuffer to... wait for it... *1 sec* - based on how
> badly LTE was performing for some people
>
> so if you control your own videoconferencing tools, collecting more
> metrics there would be usef
>
> One very interesting metric a netflix streamer could be collecting is
> differences in tcp RTT (assuming you slowed traffic down in europe,
> especially, to a lower quality?), hour by hour, day by day.
>
> another one is packet loss... retransmits...
>
>
>
>> SERGEY FEDOROV
>>
>> Director of Engineering
>>
>> sfedorov@netflix.com
>>
>> 121 Albright Way | Los Gatos, CA 95032
>>
>>
>>
>> On Thu, Apr 23, 2020 at 1:00 PM Jonathan Foulkes <jf@jonathanfoulkes.com>
>> wrote:
>>
>>> Confirmed, and I go there all time as well, you’d think it would be the
>>> first thing Google would show us.
>>>
>>> At least Fast.com is on the first page, but they don’t pro-actively show
>>> latency tests, especially on the upload.
>>>
>>> BTW- I’ve suggested they support a URL request format where we can
>>> pre-set options that engage and show results for the bloat tests. This way
>>> we can share that pre-formated link and the users who click on it
>>> immediately see a bloat metric.
>>> Maybe if a few more suggest this as well, it will climb in priority.
>>>
>>> - Jonathan
>>>
>>> > On Apr 23, 2020, at 3:38 PM, Dave Taht <dave.taht@gmail.com> wrote:
>>> >
>>> > dslreports.com is only on the third page of the search results.
>>> >
>>> > https://www.google.com/search?q=internet+speed+test
>>> > _______________________________________________
>>> > Bloat mailing list
>>> > Bloat@lists.bufferbloat.net
>>> > https://lists.bufferbloat.net/listinfo/bloat
>>>
>>> _______________________________________________
>>> Bloat mailing list
>>> Bloat@lists.bufferbloat.net
>>> https://lists.bufferbloat.net/listinfo/bloat
>>>
>>
>
> --
> Make Music, Not War
>
> Dave Täht
> CTO, TekLibre, LLC
> http://www.teklibre.com
> Tel: 1-831-435-0729
>

[-- Attachment #2: Type: text/html, Size: 8501 bytes --]