Workload diagnosis
Use waits, latency, throughput, blocking, CPU, memory, storage, and concurrency evidence to define the bottleneck.
← All practice areas
Tune from evidence
Optimization and performance
Performance tuning should begin with a measurable problem. Work from workload history, waits, resource pressure, execution plans, and query behavior to identify the limiting component before changing SQL, indexes, configuration, or infrastructure.
Core focus
What this practice covers
Query and plan tuning
Analyze execution plans, estimates, access methods, parameter sensitivity, and regressions before rewriting code.
Sustainable improvements
Balance index benefits against write cost, verify changes under realistic load, and monitor for regression.
Typical work
What teams work on
- SQL Server Query Store and PostgreSQL pg_stat_statements analysis
- Execution-plan and EXPLAIN review
- Index design, consolidation, and maintenance
- Blocking, deadlock, and concurrency investigation
- Parameter sensitivity and plan stability
- Performance baselines, load testing, and capacity forecasting
Results
What good looks like
- A measured baseline and verified improvement
- Faster queries without unnecessary indexes
- Lower operational risk during tuning
- A monitoring trail that catches future regressions