A lot has been written about backhaul links, NStreme and bandwidth testing, and tonight I made a lab test to see how far I could push a couple of RIC/522. Later on they will be depoyed as a backhaul in real life.
In fact I’m not impressed by the results, so I’m asking everybody with skills to comment on my setup.
Pair of RIC/522 with RB333 and R52H
5GHz turbo, manual TxPower
NStreme with polling, Disable CSMA, Framer policy=best fit, Framer limit=3200
Disabled all unneeded packages
Disabled Conntracking
I set up an EOIP tunnel according to this description. Only difference is that I had to config one side as station instead of AP bridge, otherwise I got no connection.
The BW test was performed using one notebook with Mikrotik BTest at each end, to avoid RB CPU consumption for generating test data.
CPU load of the RB333s were about 60-70% at the most.
Signal level around -55 dBm on both sides, clear channels, but I could not pass more than about 18/10 Mbps when going 2-way, about 30-35 when running one direction at a time.
Can the somewhat lousy results be explained by the test being run over EOIP (I’ve read somewhere that this gives a performance hit), or are there “holes” in my config that prevents me from getting further?
I know Nstreme dual performs better, but at this stage I’m going single.
Something I forgot?
