The discussion about the new hAP AX S router and speeds prompted me to look at what kind of speeds I’m getting on my LAN on Wi-Fi 6. I’ve “only” got a 300Mbps Virgin Media connection and I indeed get nearly those speeds on Wi-Fi 6 on my PC.
What’s the easiest way to carry out speed tests between my hAP ax 2 router and my PC? I know about Tools, Speed Test on the router but isn’t that some RouterOS proprietary test between Mikrotik devices? (I’ve run it between my hAP ax 2 and hAP ax lite).
Am I getting into some use of containers? I’ve never tried those on RouterOS. I see that Docker is a container. Is it feasible to run networkstatic/iperf3 - Docker Image | Docker Hub on the hAP ax 2 and my PC?
I prefer to run iPerf on "bare metal", meaning having a "server" and a "client" on two different machines. One machine is attached to the switch through wire, the client is connected wireless (when doing wireless speed tests).
No experience with running docker on a MikroTik, therefor no clue if it influences the performance.
AX2 doesn't have USB interface so I wouldn't use it for running containers.
It can be done but I doubt it's wise to do so ...
I have run docker-iperf containers already from AX3, RB5009 and Synology NAS.
All cases I got max speed (read: 950Mbps-ish using TCP with multiple streams).
If you don't have separate devices for iperf, speedtest can be used but you need to be careful.
Major attention point here: always go through the segment you want to test. Never terminate on it.
Otherwise you will load that device with AND speedtest (VERY resource heavy) AND network tasks which will without any doubt negatively impact your test results.
Also keep an eye on CPU for both devices being used for this when using speed test. The moment one of them reaches 100%, your results are to be considered faulty.
The ability to choose different servers in the same country or different country ISP to ISP, will give one a decent sense of throughput you can expect.
If you want to test your WIFI, just use your iphone and connect to ookla and compare local ISP to local ISP and so on. Clearly your ISP is the limiting factor in your case as 300Mbps is easily outpaced by wifi 6/6e/7 so your problem is not your wifi LOL.
I do measure the throughput of CAP by running an iperf3 container on capsman. Then run iperf3 test from e.g. Android smartphone. Even for wifi-qcom-ac the values are between 500-600mbit/s when wireless device is in clear sight to CAP.
Many speedtest programs test only one direction at a time. First they do a download, measure that, then they do an upload, measure that, and present the results.
When you want to know the speeds of simultaneous uploads and downloads, it is more difficult to find a solution. Of course the MikroTik Tools→Bandwidth Test can do that, and when your client system is Windows you can find a client on https://mikrotik.com/download (look at the bottom of the page).
If you have Windows, btest.exe is the easiest, IMO. If you want to construct a real test of the LAN speeds, you want to run the both server and client (iPerf or RouterOS/btest.exe) outside the systems involved.
The underlying issue is the CPU is used by running any bandwidth test server/client on the router you're measuring, since processing bandwidth test is not same impact as routing/switching.
Hence running speedtest (ookla) might be a test of JAVA performance on client device (if using ookla GUI app) or browser performance (if using a browser on client device) ... can give pretty lame results if client device app/browser is not really optimized (or uses too much CPU to display real-time happenings).
So either use iperf or ookla speedtest CLI (yes, there is one for linux), both are easiest on testing devices with regard to UI rendering.
As I said, all fake, as it doesnt matter what iperf says, aint gonna be what the OP sees at his desktop.
If you want to be a nerd and know what the max capability of your 'metal' then by all means.
Just being a sheite disturber.
It helps to be sure where the choke point is of your network.
For most it will effectively be ISP connection.
But for some it might be a stupid switch only capable of 100Mb connections, or a faulty cable ... you can only know when you test.
It also helps to be sure your internal streams can flow at the speed you intend them to run.
E.g. when copying data from NAS to PC.
No Ookla test is going to tell you that. ISP is not involved there.
I have the feel that ookla results do not reflect every day real world throughput. ISP know how ookla speedtest traffic looks like, so the can prioritize. Aside that, the topic is about local network throughput testing.
I read the first sentence, where the chap talked about lan speeds and then talked about his ISP throughput, and wifi, so thought that lan wan was on the table................ Yes if just devices on the network ( switch to switch thru the router etc,) then nothing to do with ookla.
All I wanted to do was ascertain that Wi-Fi 6 was running at full-speed between the hAP ax2 and my desktop PC which iperf3 has shown it is. My first thought made me think about running iperf3 on the hAP ax2 but as pointed out, I didn’t need to do that.
The scenario is:
Server → 1Gbps wired → Netgear switch → 1Gbps wired → Desktop = speed is 1.04Mbit/s
Server → 1Gbps wired → Netgear switch → 1Gbps wired → hAP ax 2 → Wi-Fi 6 → Desktop = speed is same
So this infers that the bottleneck here are the 1Gbps components and that Wi-Fi is running very nicely.
Fully appreciate that real world throughput will be lower.
