@phinn In the interest of maintaining up-to-date info in the ever evolving landscape of OpenWrt (which is why I love it!), your post made me go back and revisit the statement I made some time back about AQL being impacted by WED.
I run with WED enabled on my three MT6000s. However, I have come to believe, based on recent testing, that AQL may still come into play even with WED enabled.
Here's some data to chew on. I ran multiple crusader tests with varying aql_txq_limit values with WED enabled. Here are the results*:
500 - WED Enabled:
1500 - WED Enabled:
2500 - WED Enabled:
5000 - WED Enabled:
15000 - WED Enabled:
*Before anybody gets too torqued by the bandwidth numbers, I was testing at a distance from the AP (MT6000) I was tweaking without the intent of proving maximum bandwidth capability. This test was just to show effects of modifying aql_txq_limit with WED enabled while testing from a static location for each testing iteration.
So what are the thoughts of this collective think tank here? Based on these results, do you see a noticeable effect to AQL with WED enabled?