I’ve seen posts in the forum regarding similar topics as this, none of which I found had a relevant answer to my problem.
I’m currently trying to give internet to one of my clients and I seem to be having a problem. The bandwidth I’m trying to give them is 50Mbps which seemed like an easy task but is proving rather difficult now..
Using a mikrotik bandwidth test set to TCP I get a speed of about 70Mbps download and 90Mbps upload speed from the radio (MikroTik LHG XL) installed at the client.. The client receives traffic via a point to point link which receives traffic from a high site switch (CCS326) this switch receives data from a back bone (QRT5) which is linked to a switch (RB3011) from the 3011 traffic goed to a 60Ghz link and then to a cloud core to a fibre outbreak.. That’s pretty much the detail. The fibre pipe is 250Mbps currently and the Btest is from the cloud core all the way tot the LHG at the client and 70Mbps download and 90Mbps upload is obtained.
Everything seems to be great and working 100% but speedtest.net shows speeds of only 35Mbps how accurate is the Btest? Or am I missing a setting or something that is preventing me from hitting my 50Mbps at the client (the 35Mbps is when plugged into the LHG directly)
The NAT type is masquerade and I use bridges to route the traffic..
Yes I see the same issue pop up more and more across my network.. I get Btests of 50Mbps but the client only goes up to 20Mbps.. It’s almost half the speed most of the time..
Just as a side note.. I was wondering, when my phone has bad connection to a wifi network and I put intense load on it I’ve noticed that I drop a bar sometimes.. I wonder if this issue isn’t based on the same principle? The Btest is maybe light weight and doesn’t put as much load on the network as actual internet traffic so the bandwidth is very high.. But when testing the internet the full load maybe draws to much from the wifi causing a but of a dip and hence the low rate on tests.. Shot in the dark.. But yeah, I think it kinda makes sense
Rx signal level can change when switching between idle and full load. When idle, WAP will only transmit beacons with low modulation and thus transmitting at nominal Tx power. When transferring data at maximum speed, WAP will use higher order modulations and most of radios will suffer from power-backoffs … meaning that radio is not capable of transmitting at full power when using higher order modulations. Power backoffs might be a few dB, but might be enough to notice.
But I don’t think this depends on type of application doing the data transfer … a bit is a bit, doesn’t matter if produced by bTest, speedtest or iperf.