Re: [quicwg/base-drafts] Recovery: when PTO expires and no data to send (#3286)

[ianswett <[e-mail]>]

> > This exponential backoff of PING packets is less wasteful than a TCP RTO
> 
> Do you mean TCP PTO, not RTO?

TCP does't have PTO, but the exponential backoff of RTO is most similar to PTO.
> 
> > In order to detect persistent congestion correctly, you'll likely need to send multiple PINGs in this case.
> 
> I agree the use case of persistent congestion, but if we detect persistent congestion (collapsing cwnd) by consecutive PINGs, is it over-reacting to cwnd because there is no real data sent?

In QUIC, a PING packet and a packet containing data are equivalently useful from loss detection  and congestion control perspectives, so I don't think it matters whether data is sent.


The text in Section 5.3.1 of recovery says:
  "Alternatively, instead of sending an ack-eliciting packet, the sender
   MAY mark any packets still in flight as lost.  Doing so avoids
   sending an additional packet, but increases the risk that loss is
   declared too aggressively, resulting in an unnecessary rate reduction
   by the congestion controller."

So the suggestion you're describing is permitted, but it's not the recommended behavior.

-- 
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/3286#issuecomment-562585251