Speedtest slower over LAN than WiFi, but with caveats - losing my mind!

[Meeho]

The issue is somewhat convoluted so please bare with me...


General overview of the network setup:

Case 1A:
client -> LAN -> Asus RT-N66u -> wireless bridge 1 -> Mikrotik router -> wireless bridge 2 -> ISP 1

Case 1B:
client -> WiFi -> Asus RT-N66u -> wireless bridge 1 -> Mikrotik router -> wireless bridge 2 -> ISP 1

Case 1C:
client -> LAN ---------------------------> wireless bridge 1 -> Mikrotik router -> wireless bridge 2 -> ISP 1

Case 2:
client -> LAN -> Asus RT-N66u -> wireless bridge 1 -> Mikrotik router -> wireless bridge 3 -> ISP 2


ISP 1 has a 50 Mbps connection
ISP 2 has >100 Mbps connection


Speedtest results:

Case 1A: ~38 Mbps
Case 1B: ~50 Mbps (max possible)
Case 1C: ~50 Mbps (max possible)
Case 2: ~100 Mbps (capped by the 100 Mbps LAN port, not sure what the possible max is)


So, basically, when I'm connected either directly to my first wireless bridge or over the Asus on WiFi, I get the max speed. But, when connected to the LAN ports on the Asus, the speed is capped to 38 Mbps, but at the same time has no problem going over that if I'm connected to a different ISP in the backend.

Any idea what could possibly be going on here? I'm at a total loss.

 

[FNtastic]

A couple of possibilities:
- You sure the speed test server didn't change during your testing?
- happen to be using a different ethernet cable in the LAN tests?
- QoS rules in your MikroTIk router between LAN and wireless bridge 1?

 

[Meeho]

Speedtest results are repeatable over multiple servers and over days of testing.

No physical change between cases 1A and 2, other than what is after the Mikrotik router further down the network, i.e. what route the Mikrotik lets me out through.

QoS is disabled on Mikrotik, but I don't think it could distinguish between LAN/Wifi from the Asus router anyway.


One possibility that came to mind, though quite a reach, was that possibly the MTU was different out of the Asus depending on the interface and that fragmentation could be the cause, but both came out at 1500 as is set in its WAN settings (from ping packet size testing).

 

[FNtastic]

Speedtest results are repeatable over multiple servers and over days of testing.

No physical change between cases 1A and 2, other than what is after the Mikrotik router further down the network, i.e. what route the Mikrotik lets me out through.

QoS is disabled on Mikrotik, but I don't think it could distinguish between LAN/Wifi from the Asus router anyway.


One possibility that came to mind, though quite a reach, was that possibly the MTU was different out of the Asus depending on the interface and that fragmentation could be the cause, but both came out at 1500 as is set in its WAN settings (from ping packet size testing).

Maybe a long shot, but do you have another PC you could use to rule that out?

 

[Meeho]

Maybe a long shot, but do you have another PC you could use to rule that out?

Yup, tested on two PCs and two laptops. Only tested WiFi on the phone as I don't have a 802.11n capable laptop.

 

[scrappymouse]

Just curious what happens if you put your Asus-RT-N66u in AP mode and try Case 1A setup, I'd venture that the Asus is still doing some sort of routing, which if you are just using it as a switch then it doesn't need to do.

 

[Meeho]

Just curious what happens if you put your Asus-RT-N66u in AP mode and try Case 1A setup, I'd venture that the Asus is still doing some sort of routing, which if you are just using it as a switch then it doesn't need to do.

Asus is doing routing between my LAN segment and Mikrotik's LAN segment. Wireless bridge 1 is connected to Asus' WAN port.