Report: Using FreshJR QOS on main RT-AC68U (Main, RMerlin 384.11_2) + RT-AC56U (Media Bridge, john9527's 374.43_39E3j9527 LTS fork).
Additional Scripts used: amtm, Diversion, pixelserv-tls, Skynet, YazFi, ntpMerlin, connmon, scMerlin, uiDivStats on 1GB USB drive in USB 2.0 mode on the RT-AC68U.
Switched customers ISP supplied DSL modem/router to the above setup from a nominal 25Mbps download and 2.5Mbps upload speeds. The home is around 3,5ooSqFt and the area covered is over 6,000SqFt (including the driveway and backyard).
At the customer's request, I tried simply adding the Asus' setup describe above using customer's SSID's and in a double-Nat setup behind the ISP supplied modem/router. But the network experience did not improve much from the sluggish and extremely laggy one vs. just using the ISP supplied modem/router (both tested with about 25 client devices connected, including vehicles in the driveway too). Of course, I had disabled the WiFi radio on the ISP supplied modem during the test with the Asus' setup.
Next, I created new SSID's and while the customers saw improvements immediately after they had connected around 20 of their devices, I was still not happy with the network as-is. A few speed tests (from a wired GbE client, and before anyone else had connected onto the network) showed that there was a
lot of bufferbloat and the up/down speeds fluctuated wildly, each time a test was run.
Afterward, I fully and properly bridged the ISP's modem/router, created new 'secret' SSID's (so that the network was all mine for the tuning I would be doing next) and proceeded to tune Adaptive QoS while using the FreshJR script. In only 4 iterations, I had greatly improved the internet experience for my customers. Going from the worst (-) grades in dslreports.com to A+, A, A+ in a little over 60 minutes from when I had first walked in.
Creating new SSID's once more for the customers (I want this to be as 'clean' an installation as possible!), we quickly connected the RT-AC56U (Media Bridge) and the rest of the client devices. With 2 4K TV's streaming Netflix, half a dozen people on their phones (one or two using facetime) and a large download a wired computer user had initiated (GPU driver), the system still felt responsive and much faster than it ever had. With the added bonus of no one client being able to hog the bandwidth and effectively shutting out all others too!
Cutting out a lot of the advertising bandwidth used via Diversion/pixelserv-tls didn't hurt either to achieve these results.
This was just using the FreshJR QOS defaults, the only changes were made in the up/down fields in the Adaptive QoS in the GUI. The final numbers were shocking to the customers (15 down, 0.6 up), but they couldn't deny the results.
While the above may be expected results when replacing ISP supplied routers with Asus/RMerlin/john9527 powered units, it is still surprising considering that the RT-AC68U and the RT-AC56U are both over half a decade old today.
The RT-AC56U makes for a fast and stable Media Bridge and I often recommend them over PCIe solutions for desktop users, but they are becoming scarce now.
Great work by
@FreshJR,
@RMerlin,
@john9527 and all the great script-writers including
@Adamm,
@thelonelycoder,
@Jack Yaz,
@kvic and everyone else who contributes (like
@ColinTaylor and
@themiron!) that I am forgetting right now too.
All the greats above have come together to create a network environment that is so much more than just the sum of the parts.
A sincere thank you to all for giving us this level of network excellence in a way that benefits even 'mere' consumers.
(Be expecting further donations by the above customer soon).