[quicwg/base-drafts] Suggest endpoints may recognize acks after loss is declared (#3956)

[Kazuho Oku <[e-mail]>]

By design, TCP is capable of handling late-arriving acks. If a TCP endpoint receives an ack for a byte range after declaring loss of that byte range, retransmission of that byte range is cancelled.

In contrast, QUIC uses monotonically increasing packet numbers. That is definitely an improvement, but does have a downside. If loss detection is too sensitive, and if endpoints drop information regarding what had been sent in a packet at the moment that packet is declared lost, then it would not recognize the late acks.

I think that we should suggest (even if not recommend) QUIC endpoints to recognize late-arriving acks, so that our loss detection can be as aggressive and as effective as TCP, if not better.

(Please let me know if such text already exists. I am opening the issue because I could not find one)

-- 
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/3956