What exactly no of users input reflect?


What exactly the “number of users” input reflect while running a simulation.
I understand that this implies the number of users concurrently accessing the application.

I read its the load but what exactly a load mean?

Reason behind asking these basic questions:
I ran a test with different users as input (500, 800,1000, 1500, 1800, 2000) with a duration of 10 minutes and a ramp up time of 1 minute…

In all the results the mean requests/sec was around 26K-27K but the 95% response time varied from 40ms to 180 ms w.r.t to the users input.

Shouldn’t the requests/sec increase when I increase the no of users?

What behavior can I conclude from below results?

Could some one help me in understanding this?

500 Users