Hi everyone,
I got a couple of rb411 with atheros 11n on top of them, mpls bridged on two buildings 700m spaced away, free line of sight.
Both devices are inside a mimo antenna enclosure (townet 300-30-su), 20dB gain mounted in dual slant polarization.
All settings are default but for those below:
band 802.11n-only, nv2 protocol, periodic calibration disabled, HT tx/rx chains 0 and 1 enabled, nv2 tdma period size 2ms, nv2 cell radius 10 Km, nv2 security enabled.
On these conditions:
tx/rx rate is 270 Mbps,
tx/rx signal strength ch 0 is -65/-58 dbM, same values for ch 1
tx/rx ccq is 99/99 steady,
noise floor is -116 dBm,
signal to noise is 60 dB.
Above devices are connected both sides on two different switch, one 10/100, other side has 10/100/1000; routeros ethernet speed reports for both 100 Mbit full duplex.
The overall look seems good, provided the above numbers, but…
on many trials, average speed for two contemporary file transfer (1x147MB, 1x250 MB) does not exceed 23 Mbit and varies a lot.
In other words: I seen nothing like those in the 802.11n thread: my devices effective rates are quite lower.
Bandwidth test show lower performance, but my feeling is that there is a sort of ethernet traffic shaping, otherwise I cannot explain myself why network speed is so low.
A wireless scan show no interferences, during file transfer radio channel numbers varies slightly.
Is there a trick to warp real network speed, or this is the maximum that can be reached with my equipment?
Thanks for anyone who will reply.
That would probably be the max speed your RB CPU can handle. Not so sure but on http://www.routerboard.comsite you should find the throughput for your RB.
@xalexbo
I have similar configuration (RB411AH and R52nM) but my speed over this link is ~75Mbit/s.
Maybe there is a speed problem with ethernet ports on computers used to test (bad driver, etc.)
Connect these computers via ordinary switch and make file transfer speed test.
Hello, thanks for comments.
I found nothing about maximum ethernet rates that can be achieved on RBs.
Tests were done with switches connected to both sides of the devices.
Only difference, besides the connected pcs, are the switches: on the local side there is a 48 port 10/100/1000 while on the remote side there is a 24 port 10/100.
Can be related to this?
Attached here is a screen dump of a 2-way 248 Megabytes file transfer, from local to remote and from remote to local.
I think that peak rates can be seen clearly.
Comments welcome.
the limitation is your testing method. one TCP file transfer can’t go faster because of itself. You need to make several simultaneous transfers to consider this a test.
Thanks.
Putting 5 different instances of transfer remote vs. local, files from 150 to 250 MB, ethernet rate goes up to of 61 Mbit/sec peak, realistic average lies around 45 Mbit.
Under such conditions cpu is used near 50%.
Further transfers does not increases rate.
try to contact support@mikrotik.com and include the support output file.
Maybe try to do different kind of throughput testing. What results you get when you do bandwidth test through them?
