Knowing When It Is Wrong

Watch a small number of things. Connection count against the limit. Slow query log. Database size and growth rate. Replication lag if you have a replica. Index and table bloat over time.

You do not need a dashboard for a student project. You do need to know where to look, because the first time a database problem happens is a bad moment to start learning the vocabulary.

The queries that get slower are usually the ones whose data grew. A query that was fine at ten thousand rows and is unusable at ten million did not change. The plan did, because the planner chose differently as the statistics moved.

Which is why the honest habit is re-running EXPLAIN ANALYZE occasionally on real data, rather than assuming a plan you checked last year still holds.