Comparing GL-MT1300 and GL-A1300 with Wireguard, the latter experiences significantly slower speeds on the same network despite having identical configurations. Both routers are updated, set in repeater mode with the same router, use DHCP or static IP, have DDNS and a Wireguard server and employ the same Wireguard client for testing. AdGuard on/off on GL-A1300. The issue is that Wireguard on the GL-A1300 performs very slowly.
Please provide some more information about „being slow“ - about what base connection speed without VPN are we talking and about what kind of speed with VPN?
Browsing is decidedly slower on the GL-A1300: pages load immediately on the GL-MT1300 while very slowly on the GL-A1300. The browsing experience on the GL-A1300 is terrible, sometimes I have to wait several tens of seconds before the page loads completely.
Without a real speedtest (like on speedtest.net) it’s difficult to understand what’s going on.
Could be DNS issues as well.
I will perform a speed test soon, but I expect that Wireguard on the GL-A1300 should provide better or, in the worst case, similar performance to the GL-MT1300. I started with both routers in a clean default configuration and set them up in the same way, using repeater mode with DHCP and static IP, Wireguard server and forwarding the Wireguard port on the main router (using two different ports, of course). I also tried switching the port, but the results remained the same. Additionally, I used the WAN IP directly without DDNS.
Pretty sure it’s an config problem or something like this. But as I said before: Without details nobody will be able to help.
Speed test without Wireguard:
29.47 Mbps download / 9.10 Mbps upload
Speed test with Wireguard on GL-MT1300: 8.92 Mbps download / 8.59 Mbps upload
Speed test with Wireguard on GL-A1300: It cannot load the speedtest.net page.
On GL-A1300 I can browse website like Wikipedia, Google, etc but the loading is very slow. No issues on GL-MT1300.
I have always used the GL-MT1300 as a travel router without encountering any problems in various hotels, different homes and with different Internet connections. I thought I would find the same or better flexibility and performance in the GL-A1300.
Did you use a new wireguard profile or the same like on the other device? Is your DNS working?
I created new Wireguard profiles on both the GL-MT1300 and GL-A1300 routers with default settings. I used these profiles on the Wireguard client for testing. The profiles are similar and the only difference is there isn’t the Listen port on the GL-A1300 profile, section interface (auto generated in this way) and the keys. I changed the ports on both Wireguard servers and switched the ports too, but this change didn’t resolve the issue. I forwarded the necessary ports on the main router.
For the MTU settings, I adjusted the client profile for the GL-MT1300 to 1320, while the server’s default value was 1420 and this value remained consistent on both routers. For the GL-A1300 client profile, I experimented with MTU values ranging from 1280 to 1420 in one-step increments, but the issue persisted.
The DNS works and is the same on both Wireguard servers, using the default value and it works. I also tried using Google DNS and it works.
I initially used DDNS on both routers and inside the client profiles. Later, I replaced it with the WAN IP of the Wireguard servers - it is the same - to rule out any issues with DDNS.
Would not recommend changing MTUs at all. It’s not like magically make anything better.
The GL-MT1300 works very well. However, with the GL-A1300, I’ve tried various MTU values between 1280 and 1420, including the default value. Unfortunately, browsing is consistently slow on the GL-A1300. I initially used the default configuration, but after experiencing poor results, I attempted to change the MTU value without any noticeable benefits.
Yeah because it is definitely not an MTU thing. Don’t change them if you don’t know what are doing.
I know about the MTU. I am looking for a used GL-MT1300 because I need a second travel router and I will return the GL-A1300. I need a reliable travel router and, when I travel, I don’t have time to play with settings. In any case, the browsing experience of the GL-A1300 with Wireguard is not acceptable. Differently, the GL-MT1300 is plug and play and has never given me any problems on any connection.
The Slate Plus (GL-A1300) is my second most used device. The first is the Shadow, because of the low power consumption when I am in my RV without land power… That’s another story
But I can absolutely not confirm your discovery. The Slate Plus is nearly as powerful as the Slate AX and a huge improvement to the Beryl. Tested with a Fritz!Box, Debian Server and Slate AX (in my lab, behind double NAT) as endpoint (all self hosted).
You have no time to play, but set the MTU to any value without any evidence?
Lete try to explain how I determine the MTU for spooky VPN setups:
First check the fragmentation, for example via for /L %MTU in (1350,100,2050) do ping -f -l %MTU -n 2 8.8.8.8
This will send a ping with 1350byte payload and because of the ‘do not fragment’ switch, it will give an error, when the MTU is too small. After 2 trys it will do it again with +100 (1450 byte) until it reaches 2050 bytes … There you have your ‘optimal’ value. For this specific connection at this specific moment.
Second: try to find out how your ping program uses the payload. On top of the ICMP or include the IMCP header.
Thats the base speed we are talking about. Changing MTU will not make it faster or more reliable.
Changing MTU only matters, in my opinion, inside a business network where you control all network devices.
I am only comparing the performance of GL-MT1300 and GL-A1300 with their default configurations in an identical testing environment. I initially expected better performance from the latter because theoretically it should, but the former actually performs well. However, I’ve noticed that the GL-A1300 has a slower browsing experience, making it less practical for everyday use. IMHO, of course.
I found, using a similar procedure, an optimal value of MTU at 1320, and the GL-MT1300 performs very well, providing a good browsing experience. However, I cannot say the same for the GL-A1300, as its browsing performance is too slow in an identical environment.
What is the VPN service provider? I’d like to test on my side.
My measurements:
First 2 with Wireguard on, 15min delay. Second 2 with Wireguard off, 15 min delay
lupus@kira:~$ speedtest-cli
Hosted by LWLcom GmbH (Berlin) [277.19 km]: 57.521 ms
Download: 23.86 Mbit/s
Upload: 4.85 Mbit/s
lupus@kira:~$ speedtest-cli
Hosted by LWLcom GmbH (Berlin) [277.19 km]: 45.064 ms
Download: 23.70 Mbit/s
Upload: 4.48 Mbit/s
lupus@kira:~$ speedtest-cli
Hosted by KNT Internet (Riethnordhausen) [260.41 km]: 44.777 ms
Download: 25.37 Mbit/s
Upload: 4.98 Mbit/s
lupus@kira:~$ speedtest-cli
Hosted by IBH IT-Service GmbH (Dresden) [346.88 km]: 30.459 ms
Download: 25.27 Mbit/s
Upload: 5.56 Mbit/s
Hardware Setup:
Laptop - (1,5m LAN) - GL-A1300 - (2,5m WLAN) - Fritz!Box 7490 (ISP Router) - Fiber FTTH - ((Internet)) - Fritz!Box 7590 (ISP Router) - (5m LAN) - OPNSense VM - (20m LAN) - GL-AXT1800.
Software Setup:
Wireguard Server setup at the Slate AX (AXT1800), client Setup via QR Code via GL-Inet App on Slate Plus (A1300).