Hi,
We have a configuration issue with one 5GHz ptp link,
my scenario:

AP-Brige<->Station
B: 5GHz-10MHz
SS: -77/-75 dBm
NF: -102 dBm
CCQ: 88/73 %
ACK: 155 us
HW Retransmitions ratio: 1,074


When we test TCP traffic on both directions 300/300kbps is the maximum amount this link can support steady, with a couple of kbps more traffic starts to be erratic, fluctuating from 300 to n kbps.
Yesterday we conducted a lab test with the same equipment at 5 meters and we could pass 5Mbps/5Mbps. (and even more but 5/5M is the max symmetric value)

Could be an ACK timing issue?
Alignment was check and re-check several times,
any suggestions? any ideas?
This is part of a IP routed network, AP-Station is the best solution?


Thanks to all!
Edward

What hardware are you using? Routerboard? Antenna? Radio? Cables? Also what is the distance of the link?

sorry! I forgot to mention the distance!
its 20km long
I am using routerboard (433), AR5213, 29dBi solid dishes with 1 meter of RG213 on each side

What do you think?

sorry! I forgot to mention the distance!
its 20km long
I am using routerboard (433), AR5213, 29dBi solid dishes with 1 meter of RG213 on each side

What do you think?

I think you should get a hell of a lot more through it than that.

I use similar setups with SR5 or XR5 on similar distances with 29dB grid dishes and similar lengths of cable (though we use RG8). Usually can get 10Mbps TCP bi-directional without much trouble, when there are signal levels like that.

UDP uni-directional will usually hit anywhere from 15 to 25 Mbps, depending on CCQ. With your reported CCQ I would expect about 15Mbps uni-directional UDP.

perhaps a blown radio card?
any similar issue with AR5213?
suggestions?

Activate nstreme.

We have already tried NSTREAM before, no great improvements...
but, I think at least we should have a solid 3/3M tcp link as is... wouldn't be the issue at layer1-2?
I am not sure about Ack timming, anyone has tried 20km@5GHzx10MHz?

At 20km you need to use Nstream.

Ok, i will enable nstream and post pre and post results,
thanks!

Hi Edward, Nstreme cannot solve Your problem, I have a 29.5 Km 5GHz-10 MHz WDS link with two 29dBi Waveant dish, two RB411 wiht CM9 card and 5Mt of LMR400 cable (RG213 isn't good to work at 5GHz...).

Here my results:
WDS link bridge<>station WDS (no AP bridge but it's the same..)
SS: -78 dBm
CCQ: 70/75%
S/N 24 dB
Ack 120 uS
NF -102 dBm
NO Nstreme

with bandwidth test I have 8 Mbps UDP and 6/7 Mbps TCP.

I too think a blown radio card.... try to change it and update ROS....

1. If there were a blown radio card it would be obvious in signal strength.

2. My comment about Nstream is not my opinion. It is fact. 802.11b 802.11b/g was never designed for the long distances. Search around this has been discussed many many times.

Hello again folks and thanks for your feedback,
I've made the changes this weekend:
with no consumer's traffic i was able to pass 500k/500k pretty stable, as is.
then, i've enabled Nstream: 1.4M/1.4M, a bit more and it began woking erratic, in the same way it did at 600K without Nstream. so, no way to achieve 3M/3M or close, like I said earlier, no great improvements.

I insist with 3M/3M because I take another ~25Km link as reference, but no mk gear just senao and IEE 802.11, and it cost 20% of Mk,

Nstream discused many times but no conclusive answers, at least for me... maybe a little more extensive documentation would help...

RG213: true, but LMR400 is pritty hard to get here, I will try to change it at the first chance.

I will change both radios, and try again this week...
thanks!

Post your exact Nstreme settings along with all your wireless settings.

What is your signal strength on both ends?

yes, here they are:

AP Settings:
name="wlan1" mtu=1500 mac-address=00:00:00:00:00:00 arp=enabled
disable-running-check=yes interface-type=Atheros AR5213
radio-name="000000000000" mode=ap-bridge ssid="XXXXXX" area=""
frequency-mode=manual-txpower country=no_country_set antenna-gain=0
frequency=5280 band=5ghz-10mhz scan-list=default rate-set=default
supported-rates-a/g=12Mbps basic-rates-a/g=12Mbps max-station-count=2
ack-timeout=155 tx-power-mode=default noise-floor-threshold=default
periodic-calibration=enabled periodic-calibration-interval=60
burst-time=disabled dfs-mode=none antenna-mode=ant-a wds-mode=disabled
wds-default-bridge=none wds-default-cost=100 wds-cost-range=50-150
wds-ignore-ssid=no update-stats-interval=disabled
default-authentication=no default-forwarding=no default-ap-tx-limit=0
default-client-tx-limit=0 proprietary-extensions=post-2.9.25
wmm-support=enabled hide-ssid=no security-profile=default
disconnect-timeout=3s on-fail-retry-time=100ms preamble-mode=short
compression=no allow-sharedkey=no
station-bridge-clone-mac=00:00:00:00:00:00 hw-retries=3
frame-lifetime=0 adaptive-noise-immunity=ap-and-client-mode

name="wlan1" enable-nstreme=yes enable-polling=yes disable-csma=yes framer-policy=exact-size framer-limit=1000

The other end is Station, similar settings,

Signal levels are the same as posted before, even with traffic (CCQ just drop a little, about 10-20%)
One of the things that I've noticed is that latency is not constant, without traffic it oscilates from 3 to 100 ms, and with 2M/2M about 250ms, is it normal?

thanks!

Change your hardware retries back to 15.

With using ANI and Nstreme keep an eye on CPU utilization.

If this is a transparent link turn off connection tracking on the firewall

Switch your Nstream settings to framer-policy=none and framer-limit to 2000

Switch your ACK back to dynamic

I would drop down to 5mhz - you should still easily get 3/3megs

Erratic latency with low traffic on a Nstreme link is normal

You might also want to try and find your optimal data rates by manually selecting them.

supported-rates-a/g=12Mbps basic-rates-a/g=12Mbps

He's already doing that

You should check your setup again in the lab. You told us you got 5M fullduplex at your lab- setup. In your setup you nailed the bandwidth to 12Mbps in the air. With that you can't get 5M fullduplex. If you din't nailed your bandwidth in the lab-setup you are driving the link in the lab with a to high rx-level. This will overload your receiver.

What will happen if you switch your channel back to 20 MHz. Any change?

With a -77 signal he can't afford to switch to 20mhz. Geeze, it is simple math.

jwcn,
I've made the changes you suggested, It's working ok now, I can reach 3/3, It should probably do more but it is ok for now, i will start testing Nstream in a non-production environment.

erratic latency wouldn't affect voice over ip communications?

It would be great if we could have more detailed documentation about Nstream!! Is it still 802.11?

(at the lab we do not use manual data rates, nor default power, we have to simulate -77dbm in order to do a accurate test, but it was just as reference, I don't have a gear to delay 200us radio waves and simulate the 20km)

thanks a lot!