You need to place load on the server manually or through automation tools e.g. through Mercury LoadRunner.
On linux based servers, you would need to write vmstat or top on the console and then monitor different statistics of the server, e.g. CPU usage, services running on the server, threads, page faults, etc. The more you increase the load on the server, the more realistic statistics you will receive in order to determine the number of users your server can support.
On Windows server, you can use the Task Manager to perform the same operation.
You need to study performance testing for more details.