Tips for running a Server.Load test

Keep the following considerations in mind when you run a Server.Load test.

Procedure

  1. Consider the number of simulated users you plan to assign to the SUT and evaluate how that number relates to system limitations, such as disk space and memory. Server.Load creates one thread per simulated Notes® user. If, for example, you assign 100 users to one client system, 100 threads will run the test script. Note that all threads run the specified test concurrently.
  2. Set the Server.Load test parameter Thread Creation Interval parameter to stagger the creation of each user. For example, a value of 2 staggers the creation of each user by 2 seconds.
  3. Plan to enter values for the Starting Thread No. and Max No. of Users parameters. The values you enter depend on how many client systems and database users the test is simulating. For example, to simulate 400 database users across 4 client systems, with 100 users spread across the 4 clients specify these values when you run the test on each client.
    Table 1. Maximum number of users and Strating Thread number parameters

    Client

    Max. No. of Users

    Starting Thread No.

    1.00

    100.00

    1.00

    2.00

    100.00

    101.00

    3.00

    100.00

    201.00

    4.00

    100.00

    301.00

  4. Simulate the behavior of actual users by providing pauses between commands in your script. Use the built-in scripts as a reference point.
  5. Be aware of both ramp-up and steady state. Ramp-up state occurs after all threads run at least one iteration of the script. Steady state represents the server's true, sustainable performance with reproducible results. Steady state occurs when the number of Notes users on the server is equal to the total simulated users across all clients.