Ah, for that, I often use a UDP traffic generator set to 110% of "supported" bandwidth. Then, I "see" if far side obtains 100%. Ditto for a converse test.
However, doing that is fairly easy on a private network, i.e. when you have access to both sides. This can even be done, between your sites, connected to the internet.
If, as I suspect is your case, you may only have one Internet connection, and no "far side" you can use . . .
Then, often, the "best" approach I've found is using "public" Internet performance test sites. The will, generally, (should) show bandwidth rates that you expect.
Whatever kind of this testing you do, keep in mind, it can "crush" concurrent traffic performance, also concurrent traffic can "deflate" measured throughput rates, and when using public performance testing, concurrent traffic "deflation" of your performance might be due to congestion elsewhere in the Internet or the public test site, at the far end.
That said, doing multiple tests, in the "wee" hours, across different days, using different public test sites, will often provide an indication whether your ISP is actually delivering the bandwidth you're paying for.
