Could have been my words!Yeah, but one of these "other vendors" show us perfectly that the most important part here is AP side. They reuse existing old SXTs or NS5's and make users happy for a time being with speeds up to 30-50M per client and with sectors toping at 120-150M without visiting their households. They simply browse and don't feel any shortage of bandwidth. Even IPTV/OTT became possible for private houses. Replacing equipment at AP sites are far more simplier and cheaper than visiting thousands of customers and buying them new CPEs. Even gradually replacing CPEs to other vendor wouldn't be a problem, if there only were enough spectrum to operate both brands.
We've already started slowly moving to different eq, but still have thousands of SXT's (Lite5, ac, sq), SEXTANT's, QRTs, DynaDishes operating in our network. I still thank MikroTik very much for keeping their legacy wireless usable that long. And if they somehow miraculously developed something similar to other vendors AP-side equipment with similar capabilities, RouterOS alone would hold their loyal base from certain WISPs. I simply can't stand these web based GUI's or CLI's from UBNT, ePMP, etc.. and I like having unified solutions with environment I'm familiar with. So MT, please hire some workforce, reverse engineer some products in the market and revive the very reason you've once risen to where you are now - wireless.
And back to the topic
I've done a 28 client P2MP network yesterday. Overal AP's throughput doubled, single clients had higher throughput and two really poor performing client were made to 'moderate performing' clients. What did I do? Upgrade ROS to 6.42rc49, upgrade firmware, set fixed download ratio 75/25 and set hw-retries to 3 (always used 15...) So far on the combined upgrade/tweaking good result!And by the way, the introduction of TDD ratio setting, was the last major improvement we've witnessed recently. Dynamic 80/20 increased wireless DL traffic up to 20%. That's great, considering that it only took software update and some tweaking. Initial tests of NV2 improvements with high number of P2MP clients haven't showed anything appreciable, but I'll do some more testing.