[WIP] POC for actor-based simulation testing framework

[itstauq]

Description

simtester is a lightweight simulation testing framework designed to run concurrent actor-based scenarios. By defining custom actors (such as DataScientist) with specific actions, users can simulate real-world behaviors and test workflows at scale. Each actor performs random actions and then rests during a cooldown period, enabling realistic load testing and workflow simulations.

Benefits

• Concurrency: Simulates multiple actors running in parallel for realistic testing scenarios.
• Custom Actions: Easily define and extend actor behaviors using the @action decorator.
• Dynamic Cooldowns: Incorporates cooldown periods for more accurate, real-world behavior.
• Flexible: Run simulations with any combination of actors and actions to suit different use cases.
• Efficient Testing: Helps identify potential bottlenecks in workflows by modeling concurrent user behaviors.

Instructions

Follow these steps to get started:

1. Install the Package
   From the root directory of the repo, run the following command to install simtester in editable mode:

   pip install -e 'packages/simtester'

2. Run a Simulation
   To spawn 5 instances of the DataScientist actor and 1 instance of the Admin actor, use this command:

   cd tests/scenarios/bigquery/
   simtester level_2_sim_test.py::DataScientist 5 level_2_sim_test.py::Admin 1

3. What Happens Next

   • Actors will Spawn Concurrently: Five DataScientist actors and one Admin actor will be created and start running at the same time.
   • Random Actions: Each actor will randomly select and execute one of its available actions. These are methods in the class that are marked with the @action decorator.
   • Cooldown: After performing an action, each actor will enter a cooldown period, which is randomly selected between 1 to 5 seconds, as defined in the cooldown_period tuple.

4. Simulation Behavior

   • Actors continue to perform random actions and rest during cooldowns until the simulation completes.

Additional notes

• This is a POC for an alternate approach to simulation testing with easily configurable user count.
• This is in a very early stage, so feel free to reach out for any clarifications regarding the setup or running of simulations.