A 200 GB PostgreSQL database with tables bloated to three times their actual size. Autovacuum was enabled but poorly configured. How to diagnose bloat, read pg_stat_user_tables, and tune without disabling anything.
A client with Oracle 19c on Linux and disappointing performance. Default installation, no tuning. Huge Pages, semaphores, I/O scheduler, THP and security limits: everything that was missing — with the before-and-after numbers.
Friday evening, the night before a go-live. Performance collapses. Using AWR and ASH I found a hidden full table scan in a stored procedure in under ten minutes — and the production release went ahead.
A real-world PostgreSQL performance case where a LIKE '%value%' caused a full scan and degraded response times. Analysis, execution plan, and a scalable indexing strategy.
A client with a 2-billion-row transaction table and reporting queries that had gone from seconds to hours. How I solved it with Oracle partitioning — range, interval, partition pruning and local indexes.
A real-world case where the optimizer chose a nested loop on 2 million rows because statistics were stale. How to read an execution plan, spot the red flags, and fix the root cause.