Stefano Marinelli
@stefano@mastodon.bsd.cafe  ·  activity timestamp 2 weeks ago

Well, this was a surprise. Decided to run a simple LAN ping test between two of my access points, and the results were not what I expected.

Contender 1: The "mighty" Ubiquiti UniFi AC Lite
Contender 2: The tiny MikroTik mAP Lite

The winner? By a long shot, the MikroTik! It showed much better stability and lower latency.

The Stats:

- MikroTik: Avg 8.2ms / Jitter 10ms

- Ubiquiti: Avg 15.4ms / Jitter 32.1ms

Goes to show that size isn't everything in the world of networking.
Andrew Radke
@aradke@aus.social replied  ·  activity timestamp last week
@stefano the ancient, tiny, underpowered and utterly reliable mAP lite. It chews almost no power as a bonus.

And of course you can also run OSPF, BGP and MPLS on it and act the same as the big routers.

On a side note, try the wAP ax. You won’t regret it for a second. Those things absolutely rock. They are so good I’ve been told that anytime I’m touching something that could be replaced with one to just do it.
Morgan
@cereal_cable@mastodon.social replied  ·  activity timestamp 2 weeks ago
@stefano I'd be curious to know more about your test setup. Specifically what nodes you used, what they their destination pings were and the details of each node. Same for of course the access points.

I wouldn't be surprised to find Ubiquiti APs aren't all that impressive though. They really seem to be trying to do a lot more than most people even wish to do. I just want a set of APs that can support band steering and multiple SSIDs and that's about it.
Stefano Marinelli
@stefano@mastodon.bsd.cafe replied  ·  activity timestamp 2 weeks ago
@cereal_cable Those are two old access points that I had around. I upgraded them to the last firmware release, then performed a factory reset. I configured them in the same way (same SSID, same channel) and put them in the same place. I'm pinging the router - which is connected to the access points via the same cable. So the test is: test one access point, unplug, replace it with the other in the same position, plug it.
I've also an AC PRO here - maybe I could try with that, too.
Morgan
@cereal_cable@mastodon.social replied  ·  activity timestamp 2 weeks ago
@stefano Nifty, I imagine the older the AP the more likely it will be to have poor performance. I'm expecting the bridging to be a simple Linux bridge on these APs without hardware acceleration of any kind. I'd be curious if release firmware would have been more performant than current firmware even. I theorize that its the CPUs that just aren't keeping up with the bridging via software (even though its through the Linux bridges).
Syntax
@sntx@mastodon.social replied  ·  activity timestamp 2 weeks ago
@stefano It might be that the MikroTik has less dependencies or virtually less computation to process the frames. I’m not familiar with the hardware, but Ubiquiti gear has a certain level of bloat (if we can say this) compared to bare metal alternatives.

A Unikernel would always run faster than a regular kernel, just because it computes fewer LOCs.