This project contains a Locust script with a simple test to measure the performance of the grpcio library. In particular, the output files contain the performance results before and after applying this change.
Tests run inside an Ubuntu Docker container (see .devcontainer) with the following resource limits:
- 8 cores
- 10GB of memory
- 2GB of swap
The underlying machine is a MacBook Pro:
- 2.3 GHz 8-Core Intel Core i9
- 32 GB of memory
Running the test script for 20s with only one "user" sending requests at a particular point in time:
locust -f locust/locustfile.py --html locust/output/<version_name>/<version_name>.html --csv locust/output//<version_name>/<version_name> -u 1 -t 20s --headless
The metrics (RPS and response times) are very similar; the change doesn't seem to affect the performance of the library.