Hi
I apologise in advance for the long post.
At the moment I'm on a mission to optimise my internet connection for gaming as it seems delayed, whereby I'm constantly getting beaten in first person shooter games, and quite frankly it's driving me mad as I actually have decent internet speeds and a low ping.
The connection used to work very well when my network setup was much simpler many years ago. I stared off on ADSL2+ and then eventually moved over to FTTP (fiber to the premises) with my current ISP. But then again, games were simpler back too then with better net-code that actually worked and far better than the stuff I'm seeing at the moment.
Unfortunately I think I've hit a bad coincidence here; games have become worse with terrible net-code, hit-detection is flaky and other people are moaning about connection issues, but my connection and setup has theoretically improved. This leaves me wondering if there is something wrong with my setup or the games and servers have honestly worsened. This is why I would like to run tests, tweak anywhere and everywhere I can to get the best connection possible, but most importantly resolve why I have unreliable multiplayer games.
I used to have the edge on many FPS games because I had the better, optimised connection when I was on ADSL2+. I played on poweline adaptors (the closest thing I could get to cable at the time,) I placed my games console in a DMZ (which I wouldn't dare do now), and found the maximum MTU for my line to get the best I could get out of my connection. I even got picked as the dedicated P2P host out of 11 other people on many occasions.
I've recently tried going back to the old FPS games such as Modern Warfare 3 on the PS3 as I used to use this game as my benchmark as I know it's the game excel on. Sadly I really do struggle with it and have done for the last couple of years. I don't know if some of the servers have been shutdown as it is a 9 year old and matchmaking is now spread further apart, or something is misconfigured in my setup.
My setup consists of the following:-
From my BT Openreach modem/ONT I have a 3 VLAN trunk carrying across two Netgear managed switches before it reaches my main OpenWrt router.
modem/ONT > 5-port switch > 24-port switch > OpenWrt router
This VLAN trunk is predominantly a down-link for my WAN connection but it is also an up-link for a second AP from the main router. I have taken that out of the equation but it hasn't made any improvement.
The 24-port switch handles all of the cabled devices and is connected back to the OpenWrt router via LAN port 1. The single LAN port has a twin VLAN trunk consisting of the private and guest LAN. They then get tagged on along with the WAN VLAN on the 24-port switch back down to the 5-port switch before they terminate; one to the modem/ONT and the two LANs to the second AP.
The first thing I'm trying to correct is the MTU to ensure I'm not getting packet loss. My 5-port switch has a set maximum MTU of 16349 and my 24-port switch has a minimum 1518 and maximum 9216 MTU. The MTU on the 5-port switch is pre-set and I can't seem to change it to anything else. In regards to the 24-port switch, I have set the maximum MTU across all 24 ports as I have devices that use jumbo frames across the fabric of the switch and are required to achieve higher SMB transfer speeds.
As my games consoles and some other devices don't have NICs that support jumbo frames, will I get conflicts having the 24-port switch MTU set to 9216 and my PS4 for example only being able to accept 1500? Currently the Sony PlayStation consoles' MTU setting is set to automatic, but setting it to manual it can only be set to 1500. In other words should the MTU at either end be set the same, or can the 24-port switch be left open to accept up to 9216 bytes and let the client negotiate the initial MTU?
What MTU should I be setting on my OpenWrt interfaces? I have read on the forums that any LAN interfaces should be set to 1500 but will this conflict with the jumbo MTU's I have set on the Netgear switch? Out of curiosity I set the MTU of my LAN interface to 9200 (the highest LuCI will allow me to set) and I lost access to the LuCI interface so I had to SSH in, revert it back and reboot.
I'm trying to run this command ping google.co.uk -f -l 1500 in Windows Command Prompt to find the maximum MTU essentially for the WAN connection as I'm pinging an external address. Before I introduced my additional switches, setup the VLANs and had the main router connected straight into the modem/ONT, I remember having 1464 + 28 which gave me a final MTU of 1492. Now with the switches and VLANs in place, after running the above command it's wanting to be set lower at 1456 + 28 to equal 1484.
All in all I think I'm getting an MTU mismatch between my 24-port switch, the clients that connect to it and the the OpenWrt router. Could someone shed some light here please?
DSLReports have officially stopped people doing speedtests due to bandwidth. When I have done tests on their site before a week or two ago I was achieving A+ on all categories. This is with about 5% taken off my download and upload speeds for SQM with cake and layer-of-cake enabled.
Just before posting this thread I stumbled upon this Netgear Knowledge base which actually explains I need to have the same MTU at either end. I just need to figure how I'm going to get around with my current setup.
https://kb.netgear.com/25091/Guidance-on-the-use-of-jumbo-frames
Any help will be much appreciated.
Many thanks