Re: [quicwg/base-drafts] It's unclear if persistent congestion is a per-PN-space property (#3939)

[Jana Iyengar <[e-mail]>]

It shouldn't be, because congestion is a network/connection-level property. That said, @marten-seemann and I were discussing this offline, and I'm now coming to terms with the fact that it's hard to do this globally across packet number spaces.  

Basically we have no current model to track packet transmissions across packet number spaces, and implementations do not typically do this either. I think this is hard to do across PN spaces... or at least, I've not been able to figure out a reliable way to do it. 

I'm now thinking that we might want to disable persistent congestion until the handshake is confirmed. That is a departure from TCP, but I might argue that is still ok. In practice, we still have a congestion window reduction that occurs on losses in this period. Also, in practice, since the handshake will involve only a few packets in flight, losses should cause the cwnd to reduce to a small enough value.

Thoughts?

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/3939#issuecomment-662806831