Poor librespeed.org results... Is this because of WireGuard tunnel?

Hi

this the speed my provider says

↓ 300 / ↑ 50 Mbit/s

I use OpenWRT with a WireGuard tunnel to a provider in the same country:

Is this slowness because of wireguard?

It could be because of:

  • The peer
  • The LibreSpeed server
  • The peer's ISP
  • Hardware/CPU of either

You provide little information; and I get 10/10 to this test on a 100 Mbps connection.

2 Likes

Try different other speedtests to eliminate slow speedtest servers.
I tried every server offered by librespeed, and librespeed reports always significantly less than http://www.dslreports.com/speedtest.

2 Likes

dslreports server gives me also slowness

which infos do you need?

I use TP-Link Archer C7 v5 hardware with OpenWRT 19.07.3

E.g....

  • have you noticed that you're not counting the true speed between your WG peer?
  • we know nothing about the WG peer's ISP connection speed (which you have to divide by 2, if its symmetrical)

(You can use this info to count. Given I receive 10/10, even when I pick a server on my ISP, I'm skeptical.)

OK..e.g. speedtest.net let's you choose (manually) hundreds (if not thousands) of servers.

2 Likes

I not understand you. You speak complicated language, please more simple.

I use the highlighted server

https://www.ivpn.net/status

How do I notice that I not counting the true speed between your WG peer?

speedtest.net is also not better

Wireguard should not slow you down that much.
Install "VPN Policy Routing" so you can bypass Wireguard for testing.

1 Like

How to configure that "VPN Policy Routing"?
My connection always goes over WG, I tried serveral things

I now again disabled/stopped it.

If you can't understand that you're connected to your WG peer, and that we don't know the speed that connection has, then I'm not sure how to help you.

It seems like you're purposely ignoring that...or that you forgot your VPN means Virtual Private Network.

Just tell us the ISP speed of the peer, not your connection!

EDIT:

If you're using VPN peers for Internet (as you appear to be with this Wireguard peer), your maximum possible connection speed can only be as fast as 50% of your peer's ISP speed.

I'm guessing your peer has max an 80-100 Mbps connection.

I don't know how to do it just that is possible.

I don't think wireguard has that much overhead.

Really need a performance test withought wireguard,

Correct, this is normal with any VPN.

How to find this out? Please post something I can copy+paste.

I need help configure it with LuCI.
I did some try+error but it never worked. Always uses WG tunnel.

If you do not control the peer or cannot ask them, you cannot know.

That means you cannot know the maximum speed; and that means your post lacks the information needed to answer you.

1 Like

How are you working that out? I'm not sure you're right here.

Okay, I believe the assumption here is that both WG endpoint only have a single interface towards the internet and a single internet access connection running on that link. For typical enduser links or even virtual servers in the cloud that seems to be the "normal" configuration:
WG1 <-> Internet <-> WG2
or for packets inside the WG tunnel
WG1 <--WG tunnel--> WG2

Now, in the described situation packet from WG1 to the real internet will travel like this:
WG1 <--WG tunnel--> WG2 <-> Internet
or
WG1 <-> Internet <-> WG2 <-> Internet
So in short each packet will travel twice on WG2's link to the internet, which means each packets consumes 200% of the expected throughput, which in essence leaves 50% of throughput as the worst case limit. Now in the best case we will see:

WG1-egress -> WG-tunnel -> WG2-ingress -> WG2-non-tunnel interface-egress -> Internet
So that each packet still consumes twice the bandwidth on WG2, but once for egress once for ingress and hence the true bandwidth cost can be hidden. But once you hace bidirectionally saturating traffic on the link, you will see the 50% best case...

1 Like