We just finished installing a new server/fiber backbone on our Canopy network last night (10Mb!), but we are seeing customers still unable to reap the benefits of our big pipe.
The customers I am having problems with are on Advantage 900 radios, with flat panel antennas. They never see speeds higher than 1Mb, even though we are pushing through a HUGE amount of b/w. I have 5.7 customers, meanwhile, who are now getting 3Mb+ for d/l speeds. What gives here??
I called Moto support, and they suggested using a TCP tweaker, but that hasn’t helped much. Efficiency is running at about 96-98% for up and down, but we just can’t get the throughput to the customer. Most of these customers are no more than 2 miles away, either.
Has anyone ever seen this or have any suggestions as to what might be done to increase our throughput? I hate to have this nice big pipe go to waste! :evil:
For your consideration, here are the link stats from one of the customer’s radios:
Stats for LUID: 5 Test Duration: 2
Downlink RATE: 940032 bps
Uplink RATE: 297472 bps
Downlink Efficiency: 100 Percent
Max Downlink Index: 96
Actual Downlink Index: 96
Expected Frag Count: 3672
Actual Frag Count: 3814
Uplink Efficiency: 98 Percent
Expected Frag Count: 1162
Actual Frag Count: 1177
Frankly, I’m stumped. If this was a system-wide problem, I could try other things to check, but some of my customers d/l speeds are screaming now… so it isn’t happening to everyone.
Anyone??
what version software?
what is your scheduling?
what do you have your bownlink data rate set to?
if using version 7 is 2x enabled?
What is your max range set to?
Running 7.07
Using CNUT 1.1
2x is enabled in 7.0.7
Max range set for 5 miles
Downlink set for 60%
Im getting 3.5 down and 1 up at 75%. What is your config in both the ap and sm? I know its alot to list but the more you can tell me the better i can help.
Config is pretty straightforward. Not using NAT, just have a private IP scheme in place. Set on 906Mhz, with no major interference in the area. Other than what I’ve already listed, everything is still set at defaults. This is very strange…
Have you tested using UDP instead of TCP? A UDP bandwidth test will flood the link in one direction or both. You’ll get a better indication of raw bandwidth as compared to TCP. If UDP shows good throughput but TCP doesn’t, then start looking for ethernet duplex mismatches. Be careful when using auto-negotiation of ethernet duplex as it’s not a 100% reliable and this has dinged my TCP performance in the past.
Since your SM-to-AP link test(s) only show 940k, I would say that the issue is isolated to a Canopy/RF issue.
Your customers are showing real world results that match what your link test shows the canopy bridge capable of. Therefore, a UDP throughput test or any other kind of test will not be terribally useful.
It has to be something in your Canopy config. Max Range? Data/Control slots? I’ll bet its something along those lines.
Another possibility if you are using software scheduling:
I had to take my 900 units to software scheduling as per the HWS flaw that Moto released recently. In SWS, the throughput is dramatically reduced and it appears that 2x is only a HWS feature. Therefore, my effective throughput to customers (at 50% dl) is at best 850k up and down. If you are running SWS too, that throughput could very well be all Canopy is capable of.