For example, let’s say I am physically in London, and I’m using a VPN server in Texas, USA and I want to test my speed. Should I use a speed test server close to my actual location (London) or close to the VPN server location (Texas)? Which will provide better accuracy
The best location would usually be near that of your VPN server, assuming you are connecting to a major site (accessed through CDN). When you request a resource through your VPN, they will connect to the nearest CDN provider.
As other users have noted, it should be noted that some VPNs may cheat with known speed tests.
Near the VPN server location, it should do that if you let it do it automatically as well. Essentially the vpn location is the last exit point for your traffic, and to accurately gauge your speed you should choose a server near your vpn location.
I find testmy.net gives better results.
Also, be aware that VPNs might compress the traffic and inflate speedtest.net’s results.
results on their are not correct for me…
For me, testmy.net tends to vary a lot due to the way they conduct the tests. Furthermore, the issue with it is that they have very few server locations, which means the result you get doesn’t always represent your full internet speed.
In that case, they’d also compress other traffic, making the Speedtest results still reflect reality.
Some ISPs (and presumably VPN providers) will prioritize Speedtest traffic though, which IS a problem.
Yes but when using a VPN you present to all websites, including Speedtest, that you are at the VPN location. It is essentially the middle man. OPs traffic is encrypted and then sent from his laptop to OPs ISP to the VPN, which then sends the request to the respective site, in this case Speedtest. To test download speed, Speedtest sends packets from its server to VPN to ISP to OP. if the server is farther away from the VPN, internet packets will naturally have to travel more distance, decreasing speed and increasing latency.
Compared to what? How are you sure? It could be that your ISP has been fooling you by prioritizing traffic to the better-known speedtest providers.
In that case, they’d also compress other traffic, making the Speedtest results still reflect reality.
Not necessarily. Depends on your use case. For instance, if you’re using it for media, it’s most certainly already compressed.
Funny how I’ve been downvoted for a true statement though. I’ll just unsubscribe.
Because I know what kind of download speeds I can get. And I use a extension/plugin for my VPN, so I also tried another browser that doesn’t have the VPN extension.
There are two types of traffic, download and upload.
For download:
Packets are sent from Speedtest server (start point) → VPN server → ISP → to OP (endpoint).
It then calculates the time for those packets to downloads and their sizes to determine the speed of the internet connection. Clearly, as can be seen from the path above, the packets do reach the OPs computer. However, to accurately measure the maximum network speed that the OP could potentially reach, OP should choose a location that minimises packet drops and ensures maximum speed everywhere in the connection. This includes Speedtest server location. If the packets from the Speedtest server location are first sent to the VPN server, then OP should choose a location closest to the VPN server to reduce any latency to that server and maximise speed. Thus, OP will get a more accurate measure of the theoretical maximum speed s/he may get.
Upload is the same thing but in reverse (OP → Speedtest). Remember that all internet traffic interacts with your VPN right before (while downloading, after) it gets to the ISP, which is when it is sent to the computer. This is to ensure that all traffic is encrypted.
Hence, it does include the start point or end point of internet traffic.
Okay yep that confirms it.
Any time I actually want an accurate speed number, I just fire up a tiny instance on AWS, generate a 1GB random file, serve it with apache, and then download it with wget.
Not exactly, no. Internet traffic either flows to the OPs physical location, or from it. However, if connected to a VPN, OPs internet does not communicate directly with anything. It does so through the VPN, hence why a server closest to the VPN server location is best. Please read my previous comments. It explains this properly, and you don’t back your arguments with much evidence or explanation
The answer to all those questions is the OP and his computer, along with his location. However, all those requests don’t affect the result of a speedtest.
because the request also has to go through the middle man before reaching speedtest. Thus, you are going to be worsening a speedtest by having it farther away from your VPN server.