Netconf discussions, part 1

far too much queueing is done inside the firmware blob where it can't be changed, or even monitored.

We saw how BQL deployment drasically improved network throughput and handling on wired interfaces, we need something similar for wireless, but wireless has the added complication that the time needed to transmit a set of bits varies drastically from station to station, and also over time when talking to the same station.

We need to be able to work inside what are currently firmware blobs to be able to measure what's happening and experiment with fixes.

As far as MU-MIMO goes (the ability to transmit to multiple stations at the same time), the performance of the existing closed-source firmware is a great example of why we need this access. In every device I've seen a review for, turning on this feature produces a fantastic improvement in the transmitted throughput, up until about 12 devices, at which point the firmware gives up and abandons the protocol entirely, dropping throughput back to a single device.