Interesting. So, either iPerfing direct from the UDM-SE to a 10Gb iPerf server, or running nPerf to a 10Gb server on a machine that has a 10Gb link to the UDM-SE, I get approx half the speeds that the Unifi built in Speedtest comes back with.
Seems a bit weird to me. I know I was getting full speeds on my clients back in October as I took a screen shot of it to show off! After that, I’m not really sure when it dumped off. Obviously the migration had an affect but don’t know if this dumped before then or has just not recovered post migration.
The discrepancy between the UDM built in Speedtest and the iPerf results directly from it are a bit strange. Being as I have a default deny rule on the UDM that blocks iPerf unless paused and I don’t need to touch it when running from the UDM, it looks like running the iPerf on there is not subject to firewall rules etc. so can’t blame any processing on there. Very strange indeed…
I don’t think I’ve ever had my UDM get full speed from its built-in speed tests. I don’t know what service or servers it uses to measure it, but I pretty much ignore it.
Same on or off - for the tests from the UDM anyway. Obvs, if running from a client then it does have a slight impact. I have never seen it take more than 100Mb off though when all signatures are enabled.
Yeah, but that’s the thing. I’m getting much better results on the built in speed test! But, running iPerf from the same UDM, I only get around half the speed. Back in Oct/Nov when I was last obsessing with speeds and before all the migration shenanigans - I was getting the same full speed results on the UDM and any clients. Something has deffo changed.
Nothing changed our end. We don’t do work on the weekends at all. We don’t have control of any routing after it leaves our network. Which site/service has increased latency?
This is not looking like an issue that is affecting anyone else. Think there could be a fault there for sure. And ONT reset would be advised when able and then we’ll take it from there
I did pull the power for about 30s - can you get much fuller than that?! Interestingly, I did just seem to suffer a routing / DNS blip - i.e. some sites worked but others timed out. All OK again now. I think there’s something going on down the line somewhere.