Skip to content

Latest commit

 

History

History
79 lines (58 loc) · 2.17 KB

README.md

File metadata and controls

79 lines (58 loc) · 2.17 KB
Raw

Database Stress simulation

This is a mock database tester and CLI program that runs a thread which generates random latency and throughput values for a set duration of seconds, while also running these threads in parallel. A summary of these threads is printed after execution has completed.

Tested on Python 3.8.

Usage

The number of threads to spawn is passed as an integer when running the program. This value must be larger than 0.

Spawn 3 stress threads in parallel and analyze the results:

python analyze.py 3

Sample output:

Average Throughput (ops/s) = 43837.33
Average Latency (ms) = 11324.67
Max Throughput (ops/s) = 91320
Max Latency (ms) = 17923
Min Throughput (ops/s) = 1903
Min Latency (ms) = 2431
Throughput 95th percentile (ops/s) = 94027.5
Latency 95th percentile (ms) = 17973.0
Total threads run = 3

pid: 47789 started at Wed Oct 21 12:17:14 2020 (UTC) and finished at Wed Oct 21 12:17:16 2020 (UTC) taking 2.12 seconds to complete
pid: 47790 started at Wed Oct 21 12:17:14 2020 (UTC) and finished at Wed Oct 21 12:17:17 2020 (UTC) taking 3.13 seconds to complete
pid: 47791 started at Wed Oct 21 12:17:14 2020 (UTC) and finished at Wed Oct 21 12:17:18 2020 (UTC) taking 4.13 seconds to complete

The stress.py program can also be run independently with no analysis, the duration of the thread is passed as an argument:

python stress.py 3

Output:

Throughput (ops/s),Latency (ms)
78433,10286
28372,19763
16316,6240

How to run the tests:

Install test requirements:

pip install -r dev-requirements.txt

Run tests:

pytest

Decisions made:

  • Apart from pytest, all imports are standard libraries for improved maintainability and less overhead.

  • Initially the program wrote to stdout in columns seperated by spaces - this was later changed to a CSV style output to simplify output parsing and to reduce any future parsing errors.

  • The summary values are rounded to two decimal places for readability.

  • The tests are comprehensive and have thorough coverage. The code is fully typed with mypy and linted with flake8:

pytest --cov-fail-under 100 --cov tests/ --cov analyze --cov stress
flake8 .
mypy tests stress.py analyze.py