Problem
There is no benchmarking or load testing capability. Performance changes (buffer pool, Gunicorn workers, caching) cannot be validated without measurable baselines. Apache Bench is single-threaded and doesn't model realistic Superset usage (async SQL queries, WebSocket polling, dashboard rendering).
Fix
Create scripts/benchmark.sh using wrk or curl loops that models a realistic Superset session: login throughput, dashboard API latency, Redis latency. The script should accept the VIP address as an argument.
This is a manual diagnostic tool — do NOT add it to CI.
Acceptance Criteria
Problem
There is no benchmarking or load testing capability. Performance changes (buffer pool, Gunicorn workers, caching) cannot be validated without measurable baselines. Apache Bench is single-threaded and doesn't model realistic Superset usage (async SQL queries, WebSocket polling, dashboard rendering).
Fix
Create
scripts/benchmark.shusingwrkorcurlloops that models a realistic Superset session: login throughput, dashboard API latency, Redis latency. The script should accept the VIP address as an argument.This is a manual diagnostic tool — do NOT add it to CI.
Acceptance Criteria
scripts/benchmark.shcreated modeling realistic Superset session (login, dashboard API, Redis latency)