Executing a Discrete Event Simulation

Note: Unless otherwise noted, the term “simulation” can mean a simulation involving only Riverbed Modeler or a cosimulation with an external system.

There are three ways to execute a discrete event simulation:

Simulations are usually executed from the GUI. The difference between using the GUI and using op_mksim or op_runsim on the command line lies principally in the type of interface the simulation presents. In some cases, shell-spawned (command line) simulations are preferable to Modeler-spawned simulations, for the following reasons:

  • Execution of the Modeler GUI consumes a large amount of memory. Running from the command line makes this extra memory available for an executing simulation.
  • Shell-spawned simulations can better support interaction with users, due to unhindered access to the shell standard input and output devices (i.e., the shell window) or to windows opened by graphics libraries linked into the simulation. Simulations executed from the GUI can also gain access to the shell output device and graphics windows, but the presence of overlapping Modeler windows can be inconvenient and confusing.

Thus, based on these considerations, the following generalizations can be made:

  • Users who want convenient access to models, quick design of simulation sequences, and full benefit of Modeler’s graphical features primarily execute and debug simulations from the GUI.
  • Users doing complex development and debugging of simulations, where shell I/O and maximum availability of computer resources is important, primarily use external simulation execution and command-line debugging.