[Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Rich Brown]

I saw this overview of the now-in-testing Wi-Fi 6 at https://www.howtogeek.com/368332/wi-fi-6-what%E2%80%99s-different-and-why-it-matters/

Its multiple MIMO streams and maybe talking to multiple devices at a time seem as if they might be outside the assumptions we use.

Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Bob McMahon]

[-- Attachment #1: Type: text/plain, Size: 866 bytes --]

I haven't thought this out but 802.11ax is very different per things like
AP based scheduling of transmits (trigger frames), assigning to RU (sub
carriers), and per STA tx power control.  I think the implementations of
these is outside the standard as well.  Those writing traffic schedulers
will mostly likely be impacted.

Bob

On Fri, Jan 24, 2020 at 6:06 AM Rich Brown <richb.hanover@gmail.com> wrote:

> I saw this overview of the now-in-testing Wi-Fi 6 at
> https://www.howtogeek.com/368332/wi-fi-6-what%E2%80%99s-different-and-why-it-matters/
>
> Its multiple MIMO streams and maybe talking to multiple devices at a time
> seem as if they might be outside the assumptions we use.
>
>
> _______________________________________________
> Make-wifi-fast mailing list
> Make-wifi-fast@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/make-wifi-fast

[-- Attachment #2: Type: text/html, Size: 1490 bytes --]

Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Dave Taht]

What I had pushed for in multiple private meetings with multiple
providers was for wifi 6 to expose a per station API to the host AP
cpu. This
neatly rids the host AP of as much need to grok MU ( I really like the
potential in MU) or per station tx rates while still giving the
smarter AP CPU the ability to schedule stations, FQ packets and manage
bufferbloat. A per station API makes gang-scheduling more feasible in
MU-MIMO. There were many other suggestions made along the way
(privately) like having a transmit or receive is almost done
interrupt, "prepare!", and I hope some of those ideas were adopted.

I'd hoped of course to attract some actual paid work on wifi 6, but so
far have only a teeny grant to look over the intel ax200 mess.

On Fri, Jan 24, 2020 at 12:49 PM Bob McMahon via Make-wifi-fast
<make-wifi-fast@lists.bufferbloat.net> wrote:
>
>
>
>
> ---------- Forwarded message ----------
> From: Bob McMahon <bob.mcmahon@broadcom.com>
> To: Rich Brown <richb.hanover@gmail.com>
> Cc: Make-Wifi-fast <make-wifi-fast@lists.bufferbloat.net>
> Bcc:
> Date: Fri, 24 Jan 2020 12:49:36 -0800
> Subject: Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?
> I haven't thought this out but 802.11ax is very different per things like AP based scheduling of transmits (trigger frames), assigning to RU (sub carriers), and per STA tx power control.  I think the implementations of these is outside the standard as well.  Those writing traffic schedulers will mostly likely be impacted.
>
> Bob
>
> On Fri, Jan 24, 2020 at 6:06 AM Rich Brown <richb.hanover@gmail.com> wrote:
>>
>> I saw this overview of the now-in-testing Wi-Fi 6 at https://www.howtogeek.com/368332/wi-fi-6-what%E2%80%99s-different-and-why-it-matters/
>>
>> Its multiple MIMO streams and maybe talking to multiple devices at a time seem as if they might be outside the assumptions we use.
>>
>>
>> _______________________________________________
>> Make-wifi-fast mailing list
>> Make-wifi-fast@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/make-wifi-fast
>
>
>
>
> ---------- Forwarded message ----------
> From: Bob McMahon via Make-wifi-fast <make-wifi-fast@lists.bufferbloat.net>
> To: Rich Brown <richb.hanover@gmail.com>
> Cc: Make-Wifi-fast <make-wifi-fast@lists.bufferbloat.net>
> Bcc:
> Date: Fri, 24 Jan 2020 12:49:53 -0800 (PST)
> Subject: Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?
> _______________________________________________
> Make-wifi-fast mailing list
> Make-wifi-fast@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/make-wifi-fast



-- 
Make Music, Not War

Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-435-0729

Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Rich Brown]

> On Jan 24, 2020, at 9:06 AM, Rich Brown <richb.hanover@gmail.com> wrote:
> 
> I saw this overview of the now-in-testing Wi-Fi 6 at https://www.howtogeek.com/368332/wi-fi-6-what%E2%80%99s-different-and-why-it-matters/
> 
> Its multiple MIMO streams and maybe talking to multiple devices at a time seem as if they might be outside the assumptions we use.

It's worse than I thought. I just watched this explainer video from ExtremeNetworks: https://www.youtube.com/watch?v=owBrkFk9afM

If I understand correctly, they want the AP to solve a hard (bin-packing) problem, in real-time, with unclear rules for maximizing client goals (should the VoIP packet go first?). And no mention of airtime fairness or latency...

Or am I missing something? Thanks.

Rich

Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Dave Taht]

Rich Brown <richb.hanover@gmail.com> writes:

>> On Jan 24, 2020, at 9:06 AM, Rich Brown <richb.hanover@gmail.com> wrote:
>> 
>> I saw this overview of the now-in-testing Wi-Fi 6 at
>> https://www.howtogeek.com/368332/wi-fi-6-what%E2%80%99s-different-and-why-it-matters/
>> 
>> Its multiple MIMO streams and maybe talking to multiple devices at a
>> time seem as if they might be outside the assumptions we use.
>
> It's worse than I thought. I just watched this explainer video from
> ExtremeNetworks: https://www.youtube.com/watch?v=owBrkFk9afM
>
> If I understand correctly, they want the AP to solve a hard
> (bin-packing) problem, in real-time, with unclear rules for maximizing
> client goals (should the VoIP packet go first?). And no mention of
> airtime fairness or latency...
>
> Or am I missing something? Thanks.

No, they punted on these things in the design. 

>
> Rich
> _______________________________________________
> Make-wifi-fast mailing list
> Make-wifi-fast@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/make-wifi-fast

Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Bob McMahon]

[-- Attachment #1: Type: text/plain, Size: 1899 bytes --]

It's part of the reasons iperf 2.0.14 has so many new latency, trip-time,
start-time, connect, etc related features. Peak avg throughput is no longer
a valid proxy for "performance."

From a testing view, attenuation or range is no longer sufficient either.
Phase shifters are needed per things like VR/AR as truly optimizing the
number of spatial streams is needed too.

The loss function(s) to be optimized (minimized) is far from trivial in
both the definition and in [re]-computing in "real-time"

WiFi engineers have more work to do.

Bob

On Thu, Jan 30, 2020 at 2:28 PM Dave Taht <dave@taht.net> wrote:

> Rich Brown <richb.hanover@gmail.com> writes:
>
> >> On Jan 24, 2020, at 9:06 AM, Rich Brown <richb.hanover@gmail.com>
> wrote:
> >>
> >> I saw this overview of the now-in-testing Wi-Fi 6 at
> >>
> https://www.howtogeek.com/368332/wi-fi-6-what%E2%80%99s-different-and-why-it-matters/
> >>
> >> Its multiple MIMO streams and maybe talking to multiple devices at a
> >> time seem as if they might be outside the assumptions we use.
> >
> > It's worse than I thought. I just watched this explainer video from
> > ExtremeNetworks: https://www.youtube.com/watch?v=owBrkFk9afM
> >
> > If I understand correctly, they want the AP to solve a hard
> > (bin-packing) problem, in real-time, with unclear rules for maximizing
> > client goals (should the VoIP packet go first?). And no mention of
> > airtime fairness or latency...
> >
> > Or am I missing something? Thanks.
>
> No, they punted on these things in the design.
>
> >
> > Rich
> > _______________________________________________
> > Make-wifi-fast mailing list
> > Make-wifi-fast@lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/make-wifi-fast
> _______________________________________________
> Make-wifi-fast mailing list
> Make-wifi-fast@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/make-wifi-fast

[-- Attachment #2: Type: text/html, Size: 3117 bytes --]

Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Dave Taht]

Are "RvR vs latency" tests part of any testing regime outside of google yet?

http://flent-newark.bufferbloat.net/~d/Airtime%20based%20queue%20limit%20for%20FQ_CoDel%20in%20wireless%20interface.pdf

On Thu, Jan 30, 2020 at 5:20 PM Bob McMahon via Make-wifi-fast
<make-wifi-fast@lists.bufferbloat.net> wrote:
>
>
>
>
> ---------- Forwarded message ----------
> From: Bob McMahon <bob.mcmahon@broadcom.com>
> To: Dave Taht <dave@taht.net>
> Cc: Rich Brown <richb.hanover@gmail.com>, Make-Wifi-fast <make-wifi-fast@lists.bufferbloat.net>
> Bcc:
> Date: Thu, 30 Jan 2020 17:20:09 -0800
> Subject: Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?
> It's part of the reasons iperf 2.0.14 has so many new latency, trip-time, start-time, connect, etc related features. Peak avg throughput is no longer a valid proxy for "performance."
>
> From a testing view, attenuation or range is no longer sufficient either.  Phase shifters are needed per things like VR/AR as truly optimizing the number of spatial streams is needed too.
>
> The loss function(s) to be optimized (minimized) is far from trivial in both the definition and in [re]-computing in "real-time"
>
> WiFi engineers have more work to do.
>
> Bob
>
> On Thu, Jan 30, 2020 at 2:28 PM Dave Taht <dave@taht.net> wrote:
>>
>> Rich Brown <richb.hanover@gmail.com> writes:
>>
>> >> On Jan 24, 2020, at 9:06 AM, Rich Brown <richb.hanover@gmail.com> wrote:
>> >>
>> >> I saw this overview of the now-in-testing Wi-Fi 6 at
>> >> https://www.howtogeek.com/368332/wi-fi-6-what%E2%80%99s-different-and-why-it-matters/
>> >>
>> >> Its multiple MIMO streams and maybe talking to multiple devices at a
>> >> time seem as if they might be outside the assumptions we use.
>> >
>> > It's worse than I thought. I just watched this explainer video from
>> > ExtremeNetworks: https://www.youtube.com/watch?v=owBrkFk9afM
>> >
>> > If I understand correctly, they want the AP to solve a hard
>> > (bin-packing) problem, in real-time, with unclear rules for maximizing
>> > client goals (should the VoIP packet go first?). And no mention of
>> > airtime fairness or latency...
>> >
>> > Or am I missing something? Thanks.
>>
>> No, they punted on these things in the design.
>>
>> >
>> > Rich
>> > _______________________________________________
>> > Make-wifi-fast mailing list
>> > Make-wifi-fast@lists.bufferbloat.net
>> > https://lists.bufferbloat.net/listinfo/make-wifi-fast
>> _______________________________________________
>> Make-wifi-fast mailing list
>> Make-wifi-fast@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/make-wifi-fast
>
>
>
>
> ---------- Forwarded message ----------
> From: Bob McMahon via Make-wifi-fast <make-wifi-fast@lists.bufferbloat.net>
> To: Dave Taht <dave@taht.net>
> Cc: Rich Brown <richb.hanover@gmail.com>, Make-Wifi-fast <make-wifi-fast@lists.bufferbloat.net>
> Bcc:
> Date: Thu, 30 Jan 2020 17:20:24 -0800 (PST)
> Subject: Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?
> _______________________________________________
> Make-wifi-fast mailing list
> Make-wifi-fast@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/make-wifi-fast



-- 
Make Music, Not War

Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-435-0729

Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does it violate?

[Bob McMahon]

[-- Attachment #1: Type: text/plain, Size: 4193 bytes --]

We (broadcom wifi) run many latency related tests though they're all
internal testing prior to customer release. Also, a suggestion is to
measure latency per the traffic of interest vs using an adjunct ping
stream. Finally, iperf 2.0.14 supports the trip-time option.  This gives
socket write to read or end/end latencies which an application will
experience.  It also supports the end/end queue depth per Little's law
(though there is a bug in the arrival rate that needs to be fixed per the
current git head.)

Bob

On Thu, Jan 30, 2020 at 5:29 PM Dave Taht <dave.taht@gmail.com> wrote:

> Are "RvR vs latency" tests part of any testing regime outside of google
> yet?
>
>
> http://flent-newark.bufferbloat.net/~d/Airtime%20based%20queue%20limit%20for%20FQ_CoDel%20in%20wireless%20interface.pdf
>
> On Thu, Jan 30, 2020 at 5:20 PM Bob McMahon via Make-wifi-fast
> <make-wifi-fast@lists.bufferbloat.net> wrote:
> >
> >
> >
> >
> > ---------- Forwarded message ----------
> > From: Bob McMahon <bob.mcmahon@broadcom.com>
> > To: Dave Taht <dave@taht.net>
> > Cc: Rich Brown <richb.hanover@gmail.com>, Make-Wifi-fast <
> make-wifi-fast@lists.bufferbloat.net>
> > Bcc:
> > Date: Thu, 30 Jan 2020 17:20:09 -0800
> > Subject: Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does
> it violate?
> > It's part of the reasons iperf 2.0.14 has so many new latency,
> trip-time, start-time, connect, etc related features. Peak avg throughput
> is no longer a valid proxy for "performance."
> >
> > From a testing view, attenuation or range is no longer sufficient
> either.  Phase shifters are needed per things like VR/AR as truly
> optimizing the number of spatial streams is needed too.
> >
> > The loss function(s) to be optimized (minimized) is far from trivial in
> both the definition and in [re]-computing in "real-time"
> >
> > WiFi engineers have more work to do.
> >
> > Bob
> >
> > On Thu, Jan 30, 2020 at 2:28 PM Dave Taht <dave@taht.net> wrote:
> >>
> >> Rich Brown <richb.hanover@gmail.com> writes:
> >>
> >> >> On Jan 24, 2020, at 9:06 AM, Rich Brown <richb.hanover@gmail.com>
> wrote:
> >> >>
> >> >> I saw this overview of the now-in-testing Wi-Fi 6 at
> >> >>
> https://www.howtogeek.com/368332/wi-fi-6-what%E2%80%99s-different-and-why-it-matters/
> >> >>
> >> >> Its multiple MIMO streams and maybe talking to multiple devices at a
> >> >> time seem as if they might be outside the assumptions we use.
> >> >
> >> > It's worse than I thought. I just watched this explainer video from
> >> > ExtremeNetworks: https://www.youtube.com/watch?v=owBrkFk9afM
> >> >
> >> > If I understand correctly, they want the AP to solve a hard
> >> > (bin-packing) problem, in real-time, with unclear rules for maximizing
> >> > client goals (should the VoIP packet go first?). And no mention of
> >> > airtime fairness or latency...
> >> >
> >> > Or am I missing something? Thanks.
> >>
> >> No, they punted on these things in the design.
> >>
> >> >
> >> > Rich
> >> > _______________________________________________
> >> > Make-wifi-fast mailing list
> >> > Make-wifi-fast@lists.bufferbloat.net
> >> > https://lists.bufferbloat.net/listinfo/make-wifi-fast
> >> _______________________________________________
> >> Make-wifi-fast mailing list
> >> Make-wifi-fast@lists.bufferbloat.net
> >> https://lists.bufferbloat.net/listinfo/make-wifi-fast
> >
> >
> >
> >
> > ---------- Forwarded message ----------
> > From: Bob McMahon via Make-wifi-fast <
> make-wifi-fast@lists.bufferbloat.net>
> > To: Dave Taht <dave@taht.net>
> > Cc: Rich Brown <richb.hanover@gmail.com>, Make-Wifi-fast <
> make-wifi-fast@lists.bufferbloat.net>
> > Bcc:
> > Date: Thu, 30 Jan 2020 17:20:24 -0800 (PST)
> > Subject: Re: [Make-wifi-fast] Wi-Fi 6 - how many of our assumptions does
> it violate?
> > _______________________________________________
> > Make-wifi-fast mailing list
> > Make-wifi-fast@lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/make-wifi-fast
>
>
>
> --
> Make Music, Not War
>
> Dave Täht
> CTO, TekLibre, LLC
> http://www.teklibre.com
> Tel: 1-831-435-0729
>

[-- Attachment #2: Type: text/html, Size: 6876 bytes --]