How do you performance test a database-heavy application?
- database performance
- query optimization
- connection pools
Answer it first, then compare. Nothing here is saved or scored.
Model answer
Focus areas: (1) Slow query identification, enable query logging and analyze during load. (2) Connection pool sizing, test with pool exhaustion scenarios. (3) Index effectiveness, monitor query plans under load. (4) Lock contention, concurrent writes to the same rows/tables. (5) Read replica lag under write-heavy load. (6) Cache hit rates, test with cold cache vs warm cache. (7) Data volume, test with production-scale data, not empty databases. Tools: database-specific profilers, pg_stat_statements for Postgres, EXPLAIN ANALYZE.
What interviewers listen for
- Workload modellingYou design the load shape from product traffic, not from tool defaults.
- Percentile literacyYou read p50/p95/p99 separately and explain why each matters.
- Bottleneck triageYou correlate metrics across application, database, and infrastructure layers.
- Capacity instinctsYou can translate a soak test into a sizing or scaling recommendation.