Community discussions

MikroTik App
 
nikola3
just joined
Topic Author
Posts: 3
Joined: Sun Jun 27, 2021 4:17 am

PTP Poor Performance Using Two Metal AC

Sun Jun 27, 2021 4:46 am

Hi all,

I've been running a 5GHz PTP long distance link using "old" Metal N devices for quite some time now.

At some point, decided to replace the N Metals with the new AC ones. I was expecting to get (more / less) the same bandwidth with the new AC Metals, when running 5GHz-onlyN.
It turns out, the new Metals are performing worse compared to the "old" ones, when running the link using 5GHz-onlyN.

Now, switching the new ones to 5GHz-onlyAC, I the bandwidth increased slightly, but still less stable, and less bandwidth compared to the old ones, running 5GHz-onlyN.

Needless no say, everything else is the same (except for the two devices which were changed).

It'd be great if anyone has an idea, or a hint, to why this is the case.

Many thanks! Nikola
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: PTP Poor Performance Using Two Metal AC

Wed Jun 30, 2021 3:13 pm

Major difference could be the physical antenna (gain), and the channel and channel width used. (Don't leave it on "auto" , don't use XXXX or XX)

To find "why it is the case" start at the values in the "Registration table".
Klembord-2.jpg
Interesting values are TX/RX signal strength, Signal to Noise, TX/RX CCQ, which cause the resulting TX Rate and RX Rate (and the fluctuation of this).
rate syntax = "'rate'Mbps-'bandwidth'MHz/'# streams'S/Guard interval" (Streams = number of antenna used)
802.11 n uses the HT-MCS rates , while ac uses VHT-MCS rates.) http://mcsindex.com/

Data throughput will be around 50% of the TX/RX Rate at max, reduced by the CCQ percentage factor, further reduced by any other transmitter (co-channel and adjacent (overlapping) channel) interference, and the usable interface rate will be slower than the slowest transmitter in that channel, unless you use nv2 protocol.
You do not have the required permissions to view the files attached to this post.
 
User avatar
rextended
Forum Guru
Forum Guru
Posts: 11967
Joined: Tue Feb 25, 2014 12:49 pm
Location: Italy
Contact:

Re: PTP Poor Performance Using Two Metal AC

Wed Jun 30, 2021 3:32 pm

Export of the config?
RouterOS ersion?

Ue only-ac on both, and 802.11 as protocol, do not use nv2/nstreme with new ac chipset
on security profile do not set wpa or tkip, only wpa2 and aes-ccm
 
nikola3
just joined
Topic Author
Posts: 3
Joined: Sun Jun 27, 2021 4:17 am

Re: PTP Poor Performance Using Two Metal AC

Wed Jun 30, 2021 10:54 pm

Thank you both for the responses and detailed info.

Regarding this PTP link, antennas haven't been changed. To make the troubleshooting a bit more focused, for know I'm focusing on 5GHz-onlyN comparison between old and new devices.

I just replaced two Metal 5SHPn with two Metal 52SHPacn devices. For testing purposes, both set of devices set to 5GHz-onlyN (since they all support it).
Single chain, using the same frequency (channel width, and all other params).

I'm including this very simple config at the bottom of the post.

The weird this is that two Metal 5SHPn devices are outperforming the new ones Metal 52SHPacn (both using 5GHz-onlyN with the same freq / config). Antennas are the same, same distance, etc.

Here is another thing I discovered while testing the new ones Metal 52SHPacn: the performance changes when I swap the AP / Client roles. Does that make sense, or I'm missing something?


P.S during tests nv2 and 802.11 n gave similar results.
# RouterOS 6.48.3

/interface bridge
add name=bridge1
/interface wireless
set [ find default-name=wlan1 ] band=5ghz-onlyn channel-width=20/40mhz-XX \
    country=no_country_set disabled=no frequency=5790 frequency-mode=\
    superchannel installation=outdoor mode=bridge nv2-cell-radius=100 ssid=\
    "A to S" wireless-protocol=nv2
/interface list
add name=WAN
add name=LAN
/interface wireless security-profiles
set [ find default=yes ] supplicant-identity=MikroTik
/interface bridge port
add bridge=bridge1 interface=ether1
add bridge=bridge1 interface=wlan1
/ip neighbor discovery-settings
set discover-interface-list=!dynamic
/interface list member
add interface=ether1 list=WAN
add interface=wlan1 list=LAN
/ip address
add address=192.168.1.205/24 interface=wlan1 network=192.168.1.0
/ip dns
set servers=8.8.8.8
/ip route
add distance=1 gateway=192.168.1.1
/system clock
set time-zone-name=Europe/Belgrade
/system identity
set name="A to S"
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: PTP Poor Performance Using Two Metal AC

Thu Jul 01, 2021 1:25 am

Well you gave us some little details about your setup. But far from enough to give any comment on your results.

Specifying MT device types is one thing already. But are old and new on the same ROS version? What antenna is used (a very important component to understand what you have as setup.)
As you speak of of a long distance link, I assume you are not using some 6dBi omnidirectional antenna.
Your configuration is one thing, but cannot be evaluated without the registration values.

Is this in Serbia?
Using "no_country_set" and "superchannel" gives you the wrong frequencies for that country. If done in Serbia, this is even illegal.
Using XX gives no control on the channel used, it could be different after every startup. Anyway in that range the channels are at center frequencies 5745,5765,5785,5805,5825
40 MHZ is (5745Ce or 5765eC), or is (5785Ce or 5805eC). Odd numbers for the European continent? But that is the way that band is defined.
Superchannel lets you chose whatever, and you just ignore any regulation or convention.
Compare : /interface wireless info country-info serbia
With /interface wireless info country-info no_country_set

Remember that the "antenna gain" must be set properly in the config. (It will reduce the EIRP TX-power to be compatible with the country regulation chosen)
The old Metal seems 2 dB stronger than the new Metal for MCS07 (28dBm versus 26dBm). But with a proper antenna gain for a long distance link, you should never need that MAX TX power, and have a clean signal. (MAX power signal mostly has already some distortion).

No registration info leaves me blind. Asymmetric performances ... could be the case. Long links will have different distortion, Fresnel zone obstructions, and interferences at both ends. And with the odd frequency choice and the higher than normal TX power some adjacent-channel interference is quite likely to disturb the receiving end. I have no idea how you estimate "the performance" for both cases.

And just gut feeling when you use the ac as n-only: driving a diesel car while gear shifting like a petrol car will never satisfy the driver.
 
nikola3
just joined
Topic Author
Posts: 3
Joined: Sun Jun 27, 2021 4:17 am

Re: PTP Poor Performance Using Two Metal AC

Thu Jul 01, 2021 1:42 am

are old and new on the same ROS version
That is correct - both on the latest stable ROS version.
What antenna is used
Will also get back with the specific antenna models, and dimensions, but they are directional, single chain, ~1.2m diameter (but will get the exact info).
Using "no_country_set" and "superchannel" gives you the wrong frequencies for that country
That is correct. These params were set for testing purposes, working setup will have the appropriate country set.
Using XX gives no control on the channel used
This is a good point - in order to test further will set a specific width, and be able to compare performance with channels being exactly the same.
when you use the ac as n-only: driving a diesel car while gear shifting like a petrol car will never satisfy the driver.
Definitely agree, this is just to be able to compare devices (since the old ones don't support ac). My reasoning was, using the same antennas, channel width, frequency and protocol, should yield similar bandwidth in the same conditions.
how you estimate "the performance" for both cases.
Running bandwidth tests, switching equipment simultaneously on both ends, during stable weather, within reasonably small time frame.

Main item I oversaw was the XX for the chan width. Will fix that and test again.

Appreciate the comments and advices!

Who is online

Users browsing this forum: Amazon [Bot], cyrq and 37 guests