I'm a PIA VPN customer and a member of their forums. We've had an ongoing forum debate with a particular member there who has repeatedly claimed in multiple threads that "speedtest.net is useless for vpn testing." He's since expanded that statement to: "Oh, and speedof.me and testmy.net are no better, no speed test sites gives a usable valid connection speed result on VPN, no matter what they use."
He argues the only way that you can speed test while on a vpn is:
Among other things he's claimed: "the speed test sites can't test back to your system while behind the VPN... If one is using a VPN service which does not stop a speed test site from discovering the path back to your actual 'system' through their VPN network then the VPN service being used is not secure. Anyone that tells you a speed test site is accurate on their VPN service is either lying to you (either purposely or out of ignorance) or the VPN service they are using is not secure and they do not know it because its impossible for speed site results to be accurate if the VPN service network is secure because it should not be able to discover the path back to your system on a secure VPN."
When pushed further on the matter he stated: "...the pings stop when using speedtest sites on VPN, those pings need to be able to get to the users computer to test path latency and on VPN they can't."
Perhaps lrryie is correct about PIA blocking pings back to the user's computer (or maybe not). But even if PIA does block pings (and I have to question that because I have no trouble using Network Utility on my Mac pinging anything I want while on the vpn), is he correct that online speed testing services, like testmy.net cannot provide any sort of reasonable speed test when on a vpn?
He argues that the one and only way to test for speed on a vpn is by uploading and downloading large files and timing with a stopwatch. I don't think most people are willing to be bothered with the inconvenience of that. I use online speed testing services like testmy.net because it's convenient. Maybe it's not as precise as what lrryie demands has to be done, but in my view it's plenty good enough. But he says it's "useless" for vpn speed testing. Is he correct? If not I would welcome any network experts here to openly challenge him in those PIA forum threads. But at the very least I'd appreciate some feedback here.
