Should I specify a different server to my ISP?

[kosymodo]

I'm after a bit advice with regard to speedtest servers. I'm noticing regular drops in my speedtest results, and have identified that these are being reported when the speedtest server being used for that test isn't my normal server.

My package is 500/500, so I set my notification thresholds to 450 Mbps download and 400 Mbps upload.

In my Environment Variables I haven't specified the speedtest server to be used. Normally, the server used belongs to my ISP. Normally, the speedtests return expected results. However, from time to time, the server used changes to one other than my ISP's server. When this happens, the results are often less than expected (sometimes as low as 12 Mbps download and 3 Mbps upload).

My question is, is speedtesting against a server hosted by my ISP 'cheating'? Am I getting more 'honest' results from the non-ISP servers?

[svenvg93]

Test against the server hosted by my ISP is not cheating. I know from professional experience that ISPs normally only "trust" their own servers as they can control the capabilities of it. Which I do understand as some servers are not build to test for certain speeds. Or the network of the other party might have congestion on certain moments which causes your speedtest results to be low.

With that said, testing with only the sever from your ISP does not give you the full picture. It only measures the speed you're getting within your ISPs network not to the rest of the internet. IMO the best test would be to test against your ISPs server and one outside their network of which you know can reach the speeds needed. You can specify the both servers to be used in the environment variables. The app will pick random any of these two to test against on a scheduled test.