Community discussions

MikroTik App
 
fleg
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 76
Joined: Mon Nov 06, 2017 12:31 pm

ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 1:39 pm

I`m finally testing ac3 on the desk (delivered over 2 month;o)) with ac2 and I ran into old problem, which I had in ac2-ac2 test...I could`t break 200Mb.
I tried different settings (NV2, diiferent channels...etc)... but the best avarege speeds was little over 200Mb.
Ok with built-in traffic manager I hade 150Mb over TCP and 250Mb over UDP, but with Iperf (2 win machine) I`ve got the best result about 210-220 Mb/s over TCP (100s test).
Where can be problem?
I had link speed 400-800Mb/s, CCQ 80-95%, I made this test on the desk (ptp) and I have the same result like ac2-ac2 before. I`m using wpa/2/psk/aes. All interfaces are bridged.
Is it true the MT is probably weak in wifi max speed compared to another products?
I`m using MT over 15 years but I never tested max throughput in wifi before becuase I don`t need it.
 
erlinden
Forum Guru
Forum Guru
Posts: 1900
Joined: Wed Jun 12, 2013 1:59 pm
Location: Netherlands

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 2:03 pm

Can you please share your config (and only use WPA2 AES, nothing else)?
Did you test with one client connected through cable and the other client wireless connected?
What are the client specs?
 
User avatar
CZFan
Forum Guru
Forum Guru
Posts: 2098
Joined: Sun Oct 09, 2016 8:25 pm
Location: South Africa, Krugersdorp (Home town of Brad Binder)
Contact:

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 2:06 pm

If both devices are on the desk, then maybe get some more space between them, as the radios are probably screaming at each other and causes noise
 
fleg
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 76
Joined: Mon Nov 06, 2017 12:31 pm

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 2:10 pm

Can you please share your config (and only use WPA2 AES, nothing else)?
Did you test with one client connected through cable and the other client wireless connected?
What are the client specs?
There is only added security profile, the rest is in default settings.
I have pc---ac3(ap mode)----wifi(ptp)-----ac2---pc.
The both PC are i5, 8GB RAM, win7 vs win10, the latest Iperf version.
Last edited by fleg on Fri Apr 09, 2021 2:20 pm, edited 3 times in total.
 
fleg
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 76
Joined: Mon Nov 06, 2017 12:31 pm

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 2:15 pm

If both devices are on the desk, then maybe get some more space between them, as the radios are probably screaming at each other and causes noise
This is my 2nd test, previous (several moths ago) was made on the another desk with same result (ac2 vs ac2).
I`ve found several threads in this forum with the same problem...it seems I`m not alone with this "ïssue".
I think I can see data rate at 788/866 and CCQ at 93% I should see more than 200Mb/s.
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 3:09 pm

Closely look at what the wireless registration tells you. You already saw the TX and RX rate, and CCQ.
But check the full line like "866Mbps-80MHz/2S/SGI" , and the signal strength (must be lower than -35dBm) as well.

In the best conditions with 866Mbps, you can expect 400Mbps payload data rate at best (high speed wifi overhead is 50%) , if only one direction is used and nobody else is using the same channels.
This gives somewhat less than 200Mbps in bidirectional traffic in the best conditions.
If the output of the ac2-ac2 link is also wifi, it is probably using the same radio, and this at least halves the data throughput. (Wifi packet with same data travels 2 times in this wifi cell)
Best conditions also means that A-MSDU and A-MPDU are left at their default value. (Without data aggregation overhead goes up to even 90% and more)
To know if somebody is in the same channels use "Freq usage" and "Snooper" to check. High signal strength does not give priority over a weaker signal.

At best condition with 400Mbps-40MHz/2S/SGI link, 99% CCQ, and ethernet connected clients, I get 290-260Mbps unidirectional traffic speed. (as expected)

93% CCQ for a desk setup is rather low. Throughput is reduced by CCQ, due to the 7% retransmits. Low CCQ will also reduce the selected interface rate.
 
User avatar
anav
Forum Guru
Forum Guru
Posts: 18959
Joined: Sun Feb 18, 2018 11:28 pm
Location: Nova Scotia, Canada
Contact:

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 3:35 pm

866/3 = 288 should be doable in LOS scenario.
 
fleg
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 76
Joined: Mon Nov 06, 2017 12:31 pm

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 11:03 pm

Sorry guys I can hear from all sides the common TP-Link or Asus can make 400-500Mb (e.g. Archer c6) and I`m not saying that UBNT or something more profi can give over 500Mb easy (Ì`m using UNIFI in my client`s sites).
Sometimes someone write he has 400mb in his RB4011, or AC2...but Google is full of reports about poor MTs wifi throughput.
 
User avatar
anav
Forum Guru
Forum Guru
Posts: 18959
Joined: Sun Feb 18, 2018 11:28 pm
Location: Nova Scotia, Canada
Contact:

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 11:44 pm

Regardless of the wifi 5 devices, advertised speeds are two way without consideration of tx loss.
So whether its tplink or MT etc........... an 866 advertised should yield around 290, if you do better great but one needs to temper expectations.
similarly advertised of 1300 should yield around 433, etc...
 
fleg
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 76
Joined: Mon Nov 06, 2017 12:31 pm

Re: ac2 vs ac3 wifi not over 200Mb

Fri Apr 09, 2021 11:57 pm

The huge problem is that I have a little over 200 with full rate (866/80/2S/SGi).
This is not normal, I`ve found another topics of angry RB4011 users.
This is not good to read because MT is my main product for networking for several years and all my network (over 100 devices) are mostly from MT`s devices (+some UBNT).
https://telekomunikace.cz/d/34391-pomala-wifi-na-rb4011
 
fleg
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 76
Joined: Mon Nov 06, 2017 12:31 pm

Re: ac2 vs ac3 wifi not over 200Mb

Sat Apr 10, 2021 12:57 am

Problem "solved".
I tried 7 beta with wifiwave2 and speed went up...180 vs 150 in TCP (Bandwith test) and 400 vs 250 in UDP.
I had upgraded only AP (AC3), AC2 is not in compatbile list.
All packages are very very in beta stage, because there is not possible to connect with station bridge mode, there is not possible to connect via settings...I had to make scan and connect every time...etc.
Guys from MT...I love your products but this is big shame on you...you probably didn`t make good implementation of 802.11ac in your radios.
 
User avatar
anav
Forum Guru
Forum Guru
Posts: 18959
Joined: Sun Feb 18, 2018 11:28 pm
Location: Nova Scotia, Canada
Contact:

Re: ac2 vs ac3 wifi not over 200Mb

Sat Apr 10, 2021 3:01 am

Buyer beware, I would have recommended the wired version of the RB4011 if you had asked.
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: ac2 vs ac3 wifi not over 200Mb

Sat Apr 10, 2021 12:14 pm

Buyer beware. Marketing numbers are a list of performances that once where tested in a lab. Numbers are correct in ideal conditions and not achievable at the same time (eg. interface rate versus range). The load does not correspond to real life protocols. (AC1200 ; sum of multiple wifi interfaces).
Buyer beware, reviews are more real life situations. But some user reviews are just inaccurate and sometimes exaggerated. They even sometimes claim performances that exceed the theory and law of physics.(*)
Mikrotik buyer beware. RouterOS allows you to set many parameters (some are still missing), but those settings are static in operation. Other brands have way fewer parameters but the operations are more dynamic and adaptive and they have some more optimising features in wifi 5. Any non-optimal configuration of the static parameters may result in poor performance.

(*)
You can calculate what your performance is compared to the theoretical possible. E.g. use the packet calculator: https://gjermundraaen.com/thewifiairtimecalculator/
Filled in for ac (VHT) MCS9, 2 spatial stream, 40MHz and 80 MHz bandwidth, a A-MPDU of 32000 bytes, 24 Mbps basic rate, "Best effort" timing (WMM), short guard interval
... the theoretical maximum uni-directional data stream with no ack required is 300Mbps and 502Mbps. With an A-MPDU of 16000bytes it becomes 240Mbps and 363Mbps
.
Klembord-2.jpg
.
A-MPDU is not a settable parameter in RouterOS, just on/off and on for priority 0 ("BE") only by default, nor is there any feedback, and the A-MSDU used in is one of the smaller sizes compared to others (ISP modem, OpenWRT, Engenius, ...) even when the A-MSDU parameter is set at the default 8192 value.
A-MPDU is a mystery with RouterOS, but is a very important parameter for performance. A-MPDU is filled with A-MSDU blocks.
viewtopic.php?f=7&t=165698#p817088

Beacon info RouterOS 802.11n
Klembord-3.jpg
.
Beacon info Others 802.11n
Klembord-4.jpg
.
Screenshots from WinFi (free wifi analyzer)

RouterOS 802.11ac
Klembord-5.jpg
.
Others like OpenWRT, Draytek, etc 802.11ac
Klembord-6.jpg
.
You do not have the required permissions to view the files attached to this post.
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: ac2 vs ac3 wifi not over 200Mb

Sat Apr 10, 2021 7:45 pm

Without the details of your setup, it's difficult what you are doing to claim this low performance.
So , weekend with rain, and covid-19 lockdown, plenty of time for a little experiment.
I have no 2nd hAp ac2 or ac3, so I have to do it with a free Omnitik ac (single CPU, weaker in performance than a hAP ac2), I used only the router not the WLAN which is not so good indoor.

As my hAP ac2 is in use, I have to test it with the current settings (5180MHz-20MHz ce) a 40 MHz wide band (max interface rate is 400Mbps). The Omnitik is on the first floor , so separation is far from ideal with the concrete floor in between, but this gives me a send speed and a receive speed of 240 Mbps.
But extrapolation is dangerous. So when nobody was using the hAP ac2 I changed the setting to (5180MHz-20MHz Ceee) a 80 MHz wide band. This is the real test.

The Omnitik is reachable over a powerline connection , so I split of the WLAN1 from the bridge on the Omnitik, and give it a different IP address.
Over that Powerline (Devolo) the Btest shows 45 Mbps. I know Btest on the routers is far from ideal, but always a bit lazy.

Receive test initiated from the Omnitik over the WLAN connection. I manipulated the Omnitik to behave as a hAP (allow indoor freq, TX power high as a hAP).
The signal is somewhat in balance. -57dBm is good enough, 49 dMn SNR, low CCQ 67%/75% is what you can expect with that floor.
The data flow is a steady 345Mbps , although the P throughput is 518 Mbps


Klembord-2.jpg
.
Now time for a send test from the Omnitik, also giving 355 Mbps
.
Klembord-3.jpg
.
Maybe it is the CPU of the Omnitik that gets overloaded, or getting too much interrupts. (80+% load, 30 000 packets/s , FP not used on transmit !?)
So when receiving in the hAp ac2 the non-FP load is still on the Omnitik.
360Mbps and 30K packets/s .... 10Kb packet size !? What does A-MSDU (=MPDU) and A-MPDU do here ????
Klembord-4.jpg
.
And one more, sending from powerful hAP ac2. Not really an improvement 400 Mbps.
Remember this setup is far from ideal ! Getting tuned results would be easy (no floor, higher basic rate, remove the 5 SSID from the hAP ac2, stop my other AP's in the same channels, ...)

hAP ac2 better than 200 Mbps: Q.E.D.

My benefit: will integrate the Omnitik WLAN as replacement of the Powerline connection. 8 times faster.
.
You do not have the required permissions to view the files attached to this post.

Who is online

Users browsing this forum: No registered users and 64 guests