When I 'speedtest' I see something what I think is pretty weird, namely:
on the device, MR8300 & VERSION="22.03.2"
...
Download: 483.56 Mbps (data used: 433.7 MB)
...
and on my clients, on average:
...
Download: 102.27 Mbit/s
...
and this is for both, wifi & eth-wire connected clients.
How to troubleshoot that - those number cannot be such a huge discrepancy, right?
Connection to my ISP is of PPPoE type, which I had - which might be the culprit here, not by itself but because there is something not "working" in oWRT/kernel/drivers - to tweak, its MTU to be specific, to get that result on the device.
I wonder if anybody here is also on PPPoE and also - which would be sheer luck - on Linksys MR8300 and... then can confirm or object my findings, then... I wonder if that might be a solid case to ask/tell devel to look into it.
Hi,
Let me ask a noobish question here, prompted by @lejeczek (thank you for your post).
I have read in different threads about some advices not to use speedtest in the device but better on the client (some people were reporting issues in terms of remaining space on the devices).
From my (perhaps too simple) point of view, it seems a good idea to have a look at the quality of the connection at the device, to avoid what you can "lose" from the device to the client. But I understand that many think it is better to check it at the client. Can you please explain why?
Thank you in advance
Because sourcing and sinking traffic (as required for both sides of a speedtest) is a different kind of load than routing packets between WAN and LAN/WLAN. That is, running a speedtest client on a router is certainly a valid test, just not one that will tell you too much about the routing performance. This is partially because the router still needs to route packets between WAN/LAN to perform the speedtest, as well as run its NAT/firewall and then in addition needs to source/sink the speedtests traffic.
Hi @moeller0
Thank you for your reply.
I would understand that it is better to stick to the client, just to have the "complete picture" of the performances. Ok, that will avoid having to try on the router (which makes my life easier, since I am not an expert at all
This really depends on what you want to measure. If the question is "up to what rate/speed is my home router good for?" I would recommend to run the speedtest client on a machine inside the home network. If the question is "in addition to routing what rate will my router be capable of serving/receiving data" then running it on the router makes sense.
Also, if you are sure your router is up to the task of running the speedtest, and you want to measure your link's speed automatically every X hours, running it on the router can be convenient as that would allow you to shut down all other machines over night. But note that you first have to check that the router's CPU is not the bottleneck.
thank you very much @moeller0
Clear. In my current setup router is defitely the bottleneck, therefore I will keep it in mind for when I decide to upgrade
(which makes my life easier, since I am not an expert at all