Re: [quicwg/base-drafts] Sender-Controlled Delayed ACK Ratio (#1978)

[janaiyengar <[e-mail]>]

I think this is basically good and doable. I agree with @siyengar that it should be a SHOULD; there is no reason to require that a peer ack every K packets, especially since it's foolish to do so if the peer receives more than K packets in one receive event.

On mechanism: if there is enough argument for doing an UPDATE_TP, that might be a way, but that's too heavy-weight for just this parameter and I don't think this parameter is compelling enough for it. OTOH, we have only two bits left (https://github.com/quicwg/wg-materials/blob/master/ietf103/octet0.pdf), which is not enough I think even for Internet environments.

How about we do a TP and then follow it up with an UPDATE_ACK_RATIO frame? This can be as small as 2 bytes in all imaginable cases, making the encoding really low-overhead.

-- 
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/1978#issuecomment-437831112