It is time to put some TCP congestion controls out to pasture

It is time to put some TCP congestion controls out to pasture

[Stephen Hemminger]

Linux supports lots of TCP congestion control options which is good for research,
but in the current world of syszbot, it makes sense to euthanize unmaintained
especially research only algorithms.

Some options on how to do this:
    - new config option TCP_CONGESTION_CONTROL_UNSAFE?
    - move to staging
    - remove completely

What is best option?

Re: It is time to put some TCP congestion controls out to pasture

[Dave Taht]

[-- Attachment #1: Type: text/plain, Size: 1045 bytes --]

Heh,  I was going to propose moving FlexIS into the kernel to spark
more new development here:

It is a new lower than best effort CC. Also LEDBAT++ would be
potentially useful.

I think you can retire bic at least. I would have to look over the
others, There are a lot of users of vegas! still! but westwood is
better... could there be an upgrade path for those still selecting old
ccs?

My understanding is BBRv3 is going to replace BBR?


On Tue, Oct 31, 2023 at 8:46 AM Stephen Hemminger
<stephen@networkplumber.org> wrote:
>
> Linux supports lots of TCP congestion control options which is good for research,
> but in the current world of syszbot, it makes sense to euthanize unmaintained
> especially research only algorithms.
>
> Some options on how to do this:
>     - new config option TCP_CONGESTION_CONTROL_UNSAFE?
>     - move to staging
>     - remove completely
>
> What is best option?
>


--
Oct 30: https://netdevconf.info/0x17/news/the-maestro-and-the-music-bof.html
Dave Täht CSO, LibreQos

[-- Attachment #2: flexis_final.pdf --]
[-- Type: application/pdf, Size: 783391 bytes --]