Re: [tsvwg] path forward on L4S issue #16

[Jonathan Morton <[e-mail]>]

> On 11 Jun, 2020, at 11:23 am, Ingemar Johansson S <[e-mail]> wrote:
> 
> Has this extended beyond the early adopter do-it-your-self community?. I
> recall from earlier in this long tread that you found off the shelf routers
> in stores, were these shipped and ready with RFC3168 support or was an
> update necessary to make it RFC3168 capable?.

Shipped and ready.  They're using essentially the same underlying mechanisms as OpenWRT, just with a more refined user configuration interface, which includes a setup wizard which walks the user through choosing appropriate shaper parameters to go with the AQM.

In fact I understand quite a few commercial retail products actually run OpenWRT off the shelf now, just with a custom user interface.  All of these implement RFC-3168, but may require hand configuration to make it effective.  I'm talking specifically about a subset of these which mention Smart Queue Management or an equivalent term on the retail packaging, and have therefore made this feature easily accessible to the end user.

As Sebastian mentioned, the IQrouter fits in this category.  It is an outstanding example which actually probes the line over a period of several days, to determine a shaping schedule which matches diurnal variations in line capacity.  Under the covers, it is OpenWRT with some customisations, running on a normal CPE wifi router.

Understand that products like this are needed only because most ISPs and modem vendors have made absolutely no effort to tackle bufferbloat at more appropriate locations in the network - not even using packet-dropping, ECN-ignorant AQMs.  I note that ISPs' voices, as opposed to those of equipment vendors belonging to one particular industry organisation, were conspicuously absent from those calling for L4S to move forward here.

 - Jonathan Morton