[SOLVED] Cause of recent unstable ping?

Status
Not open for further replies.
Jan 16, 2022
1
0
10
Hi all,

I've recently run into an issue over the last few days, where I noticed in game I was getting unstable ping (40ms-90ms fluctuations) with fluctuating packet loss. A ping test to google would verify this, showing similar fluctuations (6ms-55ms). This doesn't happen all the time though. Often, I can perform the ping test and it will only fluctuate between 6-10ms. I live in an apartment complex, and I suspect it may be that more people are using up the bandwidth now, causing the instability. Up until recently, I was able to use a VPN to bypass any throttling and instability on the network, but now even with my VPN enabled, I am getting unstable ping. I have run a speed test with and without VPN enabled. Images are attached.

I noticed my bufferbloat results looked bad, but I only use the one device on my network. Would using QoS or AQM help resolve my issues, or is this an issue strictly on the ISP side?

https://ibb.co/CBpHw4r - with VPN
https://ibb.co/6WmcFfD - without VPN

Can anyone help me figure out what's going on please? :(

Edit: I am using ethernet directly to the socket which (I think) is basically the router.
 
Solution
That is very strange that you get more speed with a VPN than without.
Generally that means there is some issue with how your ISP is connected to the internet. A overly simple fake example. Let say you are testing to a server who is using verizon as their ISP. Your ISP is cheap and only buys a small connection to verizon and buys a bigger one to ATT.
Now let say you use a VPN that has a server in ATT and that VPN provider also buys a fast connection to verizon. In that case even though the path is more complex and should be slower going via the VPN is faster.

BUT all this doesn't really matter you can't do anything about how your ISP is connected to the world all you can do is get a different ISP. In some ways...
That is very strange that you get more speed with a VPN than without.
Generally that means there is some issue with how your ISP is connected to the internet. A overly simple fake example. Let say you are testing to a server who is using verizon as their ISP. Your ISP is cheap and only buys a small connection to verizon and buys a bigger one to ATT.
Now let say you use a VPN that has a server in ATT and that VPN provider also buys a fast connection to verizon. In that case even though the path is more complex and should be slower going via the VPN is faster.

BUT all this doesn't really matter you can't do anything about how your ISP is connected to the world all you can do is get a different ISP. In some ways using a vpn service is partially getting a different ISP.

Gamers are dumb as rocks sometimes. They sit in a room and chant bufferbloat bufferbloat but have no clue what that really is.

Bufferbloat is the network equipment trying to not discard data when a connection gets overloaded. It places the data temporarily in a buffer rather than discarding it. This is actually a very good thing for every application other than online games. It actually costs money for these memory buffers so if it wasn't very
beneficial it would not exist.

Most the problem with the buffbloat tests is they are invalid. The are in effect overloading the circuit to test how much memory data buffer equipment in the path has. You only get bufferbloat when you are overloading a connection. If the connection is not being overloaded then the data is never placed in the buffers
The bufferbloat test though will force a artificial overload on even the fastest internet connection.


On large connections like yours it is highly unlikely you are overloading the bandwidth you purchase from the ISP. Any form of QoS just decides which data is left in the data buffer the longest. If there is no overload no data is being placed in the queue/buffer to begin with.

The QoS software that fixes this is fairly cpu intensive and will bottleneck most routers that support it. Most will cap your data rates under 100mbps. It works best on smaller connection say like 50mbps. It is also these smaller connection that have actual overload conditions.

Now technically there are buffers in other network equipment in the path. But you have no ability to affect the queues in those devices. You could limit your traffic but then someone else's traffic would still fill those slots in the queues.

I am not sure what your actual problem is. A game is not affected by the download speed. Most only need about 1mbps and having more does not make things run better. Ping time to some extent matter. Games need very consistent ping times BUT they also are designed to tolerate some variations.
In most games you need a spikes well over 100ms or even 200ms to be able to detect them in game. Smaller amounts are well within the design limits of the game.

Ping spikes can be caused by data being held in buffers.....which is why gamers chant "bufferbloat". Problem is if the delays are not being caused by equipment fully in your control you can do nothing about it.
 
Solution
Status
Not open for further replies.