Hi,

On Fri, Nov 21, 2025 at 09:28:31AM +0200, Lars Eggert wrote:
> Hi,
> 
> below are Cullen's raw notes from the side meeting. Thanks for letting me
> share them!
> 
> (We tried to use the "AI summary" feature of the conferencing system, but
> somehow that ended up not producing any summary, AI or not. Probably pilot
> error on my part, sorry.)
> 
> We're planning to continue the discussion in some form, please let me know if
> you'd want to be included.

Thanks Lars and Cullen for the notes, that's very useful!

I can add a point I'm not seeing there. I'm in contact with some
infrastructure people who have been facing significantly slower large
file uploads over QUIC than TCP (roughly 5-8x), to the point of having
to disable QUIC for sites using a lot of uploads. Quick tests tend to
indicate that upload performance varies a lot depending on the clients
and the server (4x) with no combination found to reach H2/TCP perf. It
is possible that this part is not as much developed in existing
components (congestion control in clients, buffering in servers), and
simply leads to QUIC just being disabled on the server side since there's
very little that can be done there to address the issue when it depends
on the client.

Just my two cents,
Willy

Reply via email to