Re: [quicwg/base-drafts] Processing buffered packets can cause hugely inflated RTT measurements (#3980)

[Kazuho Oku <[e-mail]>]

@ianswett (swapping paragraphs being quoted)
> However, one difference from the above is this doesn't rely on handshake confirmation, since this is just as much a problem for Handshake packets as it is 1-RTT packets, so I think we need to use a more general statement about the sender not knowing if the receiver has the keys to process the packets.

I am happy to see the issue addressed for all epochs if that's the preference of others.

> It also mitigates the issue for those that don't follow this new MAY.

FWIW I tend to think that we should use SHOULD, especially if we are going to also cover the delay of Handshake ACKs.

Max-ack-delay of Handshake packets are assumed to be zero. Therefore, we would see the problem on any environment where the time required for processing a TLS handshake message is relatively greater than the path delay. Including datacenter networks where RTT is small, or slow IoT devices.

-- 
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/3980#issuecomment-668848535