We have been updating now some 8 AP's and their P2MP network clients and the overall outcome is that 802.11 plain protocol with properly configured RTS/CTS ( (Relavant and varialbe settings shown only.) is working better then NV2.

We have a mix of clients getting assigned by gateway router either 7 or a 30Mbps queue.

After we've done the upgrade on both CPE's as AP we are testing the network by running a bandwidth test to at least 3 clients simultaneously.
In NV2 it is a rare event that client reaches its maximum assigned speed and the total throughput over the AP sort of 'hangs' around 40-60Mbps with the other clients testing.
The moment we'd swap to 802.11 we see speeds to client climbing up and also the throughput over the AP goes up.
When testing a single client they almost guaranteed get the maximum assigned 30Mbps (tcp traffic) and the AP's throughput sometimes more then doubles.... (not always double though, but always some improvement)

All these AP's are in a very congested spectrum environment where it is always a struggling to find some relative 'free' radio band.

we have one AP though that is a NetMetal connected to a 18dBi RF elements dome pointing towards a region where no other towers are penetrating with their signals and here we found one AP that indeed performs better in NV2!
Still I have problems to get over 100Mbps over the AP and only see some short live 'peaks' reaching 125Mbps in a 40Mhz wide channel. This unit only has 12 CPE's.

My conclusion is a repeat of some ROS versions ago; 802.11 (either 'n' or 'ac') outperforms NV2 in P2MP networks by some 10 to 100%!
We also learned some setting of one AP can never be copied into another. It always has to be tested but so far 802.11 with RTS/CTS is a very good protocol and in 'ac' we can even work with 80Mhz channel that are partially overlapping other channels due its standard (IEEE) build in interference avoiding system.

Hello !

First big thanks taking the time sharing this knowledge!

We will follow up with a test in our environment to see if 802.11 will work in our environment, which for the past 6-7 years has been using nv2.
Back in that time we used nstreme and before that 802.11 (Cisco devices), both struggled with coming and going latency, slownesses and stalled networks.
Both nv2 and nstreme was setup and tuned by help and recommendation from our distributor, the vendor and forum gurus.

We also have an ongoing test with another brand ubiquiti, at the time being it performs better than NV2 in same environment (city like), but we don't want to leave mikrotik.

I have to mention that indeed NV2 in 6.42.6 has been improved. On AP's still running NV2 we see an improvement in overall capacity after the upgrade from <6.42.4 to 6.42.6.
So in that respect MT has been right in their claim that NV2 was improved in the several 6.42.x versions.

But it doesn't let alone that we like we say in earlier v6.x.xx versions, 802.11 just performs better for us in most P2MP networks..

The only thing NV2 is better is latency. It shows a more consistent and low latency then either nstreme or 802.11. But even in the higher jitter we have on 802.11 this is still within a reasonable range and we don't provide VOIP so not that important to us.
The bulk of our clients are not using their internet for gaming neither so here again latency is not such an issue.
Like everybody around the world most of our clients heavily use any kind of communication (Skype, Wattsapp, Facetime, messenger etc) and we actually never got complaints about poor performance. Live IPTV is also used a lot. Still we don't do QOS on our network, we just make sure there is plenty of capacity at hand....

Usually we do get complaints when people are at home using their internet but don't realize they are not connected to their wifi (=our network) but instead are on G3/G4 which at times is nice, but pretty regular also sucks.... (Wining their (=our) internet is slow but when I peep in their wifi router they are not even connected!)