Re: [quicwg/base-drafts] QUIC Transport: Flow Control (#3722)

[evolodina <[e-mail]>]

> @evolodina Can you clarify what conditions you're testing, as there are are a lot of variables in flow control.

@ianswett  We used simply conditions: sender is not app-limited,  we tested FC for different bottleneck links e.g. 100Mbps, 1Gbps and different FC Limits, RTT was between 1ms und 100ms with step 10. We do not used dynamic window tuning as it’s not defined in draft as a MUST.

> So I'd expect if we did more proactive updates with better bundling, it'd only help.
> Bundling flow control updates with ACK frames makes a large amount of sense, because ACK frames free up congestion window space and window updates free up flow control.

I think it’s an interesting idea to bundle FC updates with ACK frames (if ACK must be sent every other packet). But in the case of bundling, I expect, that ackRatio parameter (e.g. equal to 10) would influence the frequently of FC-updates and thereby QUIC performance. I can implement und explore it in our implementation.

In order to have a possibility to test FC appropriate and provide comparable results, I wish, that the FC algorithm would be better defined in QUIC transport draft e.g. with a pseudocode.

-- 
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/3722#issuecomment-646088295