What the complexity buys you is that BBRs metrics max_BW, min_RTT, and the ACK aggregation/batching metrics are actual parameters of the network, and observable with passive instrumentation of the packet streams.  Traditional CC is a collection of heuristics to estimate cwnd, which has a clear interpretation in terms of action (when to send), but the optimal cwnd can't easily be observed from the packet stream.

I think this alone will have impact, in terms of being able to reason about CC behaviors.

Thanks,
--MM--
The best way to predict the future is to create it.  - Alan Kay

We must not tolerate intolerance;
       however our response must be carefully measured: 
            too strong would be hypocritical and risks spiraling out of control;
            too weak risks being mistaken for tacit approval.


On Sun, Jul 5, 2020 at 11:13 AM Jonathan Morton <chromatix99@gmail.com> wrote:
> On 5 Jul, 2020, at 9:09 pm, Stephen Hemminger <stephen@networkplumber.org> wrote:
>
> I keep wondering how BBR will respond to intermediaries that aggregate packets.
> At higher speeds, won't packet trains happen and would it not get confused
> by this? Or is its measurement interval long enough that it doesn't matter.

Up-thread, there was mention of patches related to wifi.  Aggregation is precisely one of the things that would address.  I should note that the brief description I gave glossed over a lot of fine details of BBR's implementation, which include careful filtering and conditioning of the data it gathers about the network path.

I'm not altogether a fan of such complexity.

 - Jonathan Morton

_______________________________________________
Bloat mailing list
Bloat@lists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/bloat