My primary goal is a reliable, wireless mesh backhaul and the Audience fit that void fairly well. Configuring the units in this way is cumbersome, but it works. However, the Audience is now getting long in the tooth so to speak. It's WiFi 5 and lacks similar hardware features the hAP be3 has. It's also fairly large in physical size compared to the more compact size of the hAP be3.
My question for those in the know is can the hAP be3 operate as a potential wireless mesh replacement for the Audience? On paper it sounds like that's the case, but obviously I'd like to know for sure so that I don't waste money on the Audience when a potential replacement is on its way to GA.
Here's an example of what I mean when comparing both units:
3 WiFi interfaces let you use any of them to mesh devices. It's up to you what you use them for.
Seems that 6GHz is the best choice for mesh if only it would work in you environment.
For sure, I'd probably look at using the 6GHz radios for the backhaul link and 2.4GHz/5GHz are for Wi-Fi clients. Or rather, I would imagine that's how MikroTik would suggest we do/program by default to occur. The antenna gain is the one that stands out the most since the hAP be3 (on paper) should have better signal strength.
Antenna gain of radio 3 is higher mostly due to higher frequency ... it doesn't mean that antenna is actually better. And using radio 3 as meshing backhaul might cause problems as it's in 6GHz band and won't penetrate any walls ... even in line of sight range will be lower than of lower frequency radios. All of that was implied by @BartoszP ...
I'm using eero Pro (WiFi 6) devices now in a wireless mesh and I believe that uses 6GHz on the backhaul. In theory, MikroTik configured the same way should work okay in this environment.
5Ghz or even 2.4GHz could have problems if walls are made from trendy glass.
My suggestion is based on an idea that the fastest link should be used to link/mesh devices but ... it has to be verified if it works at all or just good enough as always if it comes to WiFi domain when reality very fast verifies theory assumptions.
Generally there are two options when creating a swarm of APs with wireless used as backhaul:
true meshing where software on APs dynamically creates routing of data towards wired exit(s) ... somehow similar to xSTP. Wired ports are part of same meshing algorithm and algorithm decides how and when to use any particular wired interface as exit
"meshing" where there's static (L2) configuration of certain wifi interfaces to be used as backhaul and there is (or are a few) a dedicated AP with wired connection used as "exit"
ROS doesn't implement meshing as per bullet #1 above. I guess it could be done using some more or less smart scripts. However, MT does work nicely when configured according to bullet #2 above ... which is how "meshing" is done on Audience with using wifi2 interface (the one operating on upper half of 5GHz band with 4 chains).
So what can easily be done using any MT device with 2 or more radios is to use one of them for backhaul and the rest for serving normal stations. And as there's no "meshing" QuickSet profile (I guess there is on Audience if it's running ROS v6), one has complete freedom choosing which radio to use for backhaul. IMO there are a few criteria when deciding which radio is best suited for backhaul. Actual throughput is certainly one of them ... and here 6GHz is in principle favoured due to ability of using 320MHz wide channels (if operating BE as opposed to AX on WiFi6e) and less interference (for now) ... but that (as indicated by @BartoszP ) can not be taken for granted ... with high signal loss (due to NLoS conditions ... doesn't have to be metal wall, can be closet filled with paper) throughput can drop even below what can be achieved using narrower channels on lower frequencies. And if backhaul can't provide enough throughput, maximum rates on WiFi doesn't improve overall performance for users.
At the same time we probably want decent overlap of AP coverage for stations ... and that's usually easier to achieve with lower frequencies.
If we construct a "mesh" according to bullet #2 above (i.e. static configuration of wifi interfaces, dedicated to backhauling), APs intended to serve stations can be actually controlled and provisioned by a CAPsMAN with all of its benefits. The same possibility might exist also if "mesh" operates according to bullet #1 above.
"wireless repeater" functionality is actually a bit different (from setup and administration point of view, should be same from station point of view) ... there's a difference how "backhaul" operates (and if the "anchor AP" is actually aware that one of its "stations" is a "repeater").