Re: [quicwg/base-drafts] Should kPersistentCongestionThreshold be 2 or 3? (#2556)

[Jana Iyengar <[e-mail]>]

The TCP minRTO is 1 sec per RFC 6298, but Linux uses 200 ms if I'm not mistaken. I could be wrong, but don't think it was based on the delayed ack timer, since the delayed ack timer value is typically 200ms. (I'll confirm this.)

My high order point is that we probably do not want QUIC to declare persistent congestion (and collapse cwnd) sooner than TCP currently does in the network. Using only the PTO values might cause us to do that, and it might cause unnecessary throughput degradation.

IMO we don't lose anything by declaring persistent congestion similar to TCP. PTOs still fire the way they always do, and recovery continues the way it always does. Other machinery (such as when to call a connection or path dead for instance) can continue to use PTOs, but we just declare persistent congestion based on what we know TCP behavior to be in the wild.

-- 
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/2556#issuecomment-477684981