Hi folks,
recently i’ve upgraded a small AP based on RTL8181@100mW and 14dbi Omni antenna, working with about 10 CPE’s associations and 700 kbps average download traffic for a RB532+AR5213 (and the same omni)
after a lot of research at the forum, internet and a looot of tests i had to setup the old 8181 again…
basically everything worked great (with RB532) up to 500 kbps, more trafic and the 8ms ping for 1km link became 300ms and the 5% loss to 30%
(the setup with the 8181 has 2% loss and 30ms average between all cpe)
Have anyone had a similar issue? Could the omni be the problem? any suggestions for the next ‘upgrade’?
thanks to all
Gabriel
YES its a real headache ..
i have a link like this :
RB333+XR5+12dbi omni ====4Km====RB333+XR5+24dbi flat panel
signal strength about -70
the ping is about 2ms
when starting bandwidth test it reaches 2-3Mbps , but the ping jump to 300 !!!
its only 4 km not 304Km and nothing left to reach satellite ping 
I have the same problem …
plis somebody help …
When you run a bandwidth speed tests you are causing the CPU on both sides of the link to max out the most data out of that link which in returns takes up most of your CPU resources. This is what is causing your ping latency to go high.
Matt
I agree, bandwidth test uses all CPU, but my issue was a little bit strange, it was pc-to-internet traffic,
what do you think?
Ok so if i’m understanding you. You said that when you run a local pc to internet test over the MT links you have you get high lantency all dependent on traffic correct?
More traffic = higher latency?
Ok if i’m understanding it correctly. For me to be able to help you I will need more information.
- Do you have connection tracking turned off?
- How are your links setup (routing, eoip, wds, …)?
- Any firewall rules?
- Queues?
- Anything over normal config over than just routing…
See the bad part is with seeing your config and setup its really hard for me to see what problem your having and how to fix it.
Suggestions:
- Turn off connection tracking.
- Disable ALL unneccessary packages you don’t need (example ups and so on…)
- Check SNR (Signal to Noise Ratio)
- Check noise floor
- What your CPU usage when running your speedtest
I know that steps 3-4 on the suggestion list wont be causing your problem from what your explaining. I always double check those things anyhow just to make sure.
normally is the problem with the interference, if you wish to serve point to point. try to use panel or grid antenna. if you wish to serve area, try to use sectoral. avoid using omni because it will receive interference from all direction
Having very similar problem. Nstreme-Dual…Checked noise…floor happens when is -105 and when it’s -90. CPU usage is high and often 100% however I can test for some time and will get 20-23mbps send then recv but then start getting drop outs although the link still shows connected I can see in the traffic window where Throughput drops to 1-2mbps briefly then sometimes for a few minutes. Pings look good then latency jumps 1000 fold. Not convinced that RB333 really optimized for Nstreme-Dual as touted. Have Superchannel license and tried many many tx-rx freq permutations.