Notes for the problems that show up after launch: bad plans, awkward migrations, index debt, vacuum pressure, replica lag, and the small decisions that keep PostgreSQL, MySQL, MariaDB, and SQL Server easier to operate.
Find the queries dragging your Postgres down, read their plans, and fix them in order of impact.
Read the guide →Which indexes to add, which to drop, and how to spot redundant index debt before it costs you writes.
Read the guide →How autovacuum works, when it falls behind, and how to keep bloat from eating your storage and your buffer cache.
Read the guide →Sizing pools, PgBouncer modes, and how to stop idle-in-transaction sessions from exhausting max_connections.
Read the guide →Track lag, WAL throughput, and replication slots before they turn into stale reads or a full disk.
Read the guide →
Backups, restores, PITR, archive gaps, and recovery drills for PostgreSQL.
AWS RDS, Azure Database for PostgreSQL, Google Cloud SQL, AlloyDB, and managed PostgreSQL monitoring notes.
Connection storms, pool saturation, PgBouncer behavior, and max_connections pressure.
CPU, memory, disk, temp files, checkpoints, buffers, and workload pressure.
Index design, index debt, reindexing, constraint indexes, and evidence-driven DDL.
Lock waits, deadlocks, blocking chains, idle transactions, and transaction hygiene.
Galera clustering, InnoDB and Aria internals, replication, optimizer behavior, and MariaDB operations notes.
InnoDB internals, replication, locking, query performance, and MySQL operations notes.
HNSW, IVFFlat, recall, embedding search, and production RAG performance notes.
Replica lag, WAL growth, failover readiness, hot standby behavior, and replication slots.
Query plans, EXPLAIN ANALYZE, planner regressions, pagination, joins, and statistics.
Wait statistics, tempdb, Query Store, Always On availability groups, and SQL Server operations notes.
Autovacuum, dead tuples, bloat, wraparound, MVCC, freeze age, and visibility.
Month-end reporting pushed the OLTP pool to ninety seconds of queue depth for six hours straight, and the fix was not hardware — it was a classifier function and two resource pools. How Resource Governor actually throttles, the classifier bug…
A nightly status-update batch crossed the 5,000-lock threshold, escalated to a table lock on Orders, and blocked checkout for eleven minutes while the deadlock monitor logged nothing. How lock escalation actually decides, why it is usually a design smell, and…
An audit table created without a clustered index accumulated 40 million forwarded records over eighteen months, and its nightly scan went from four minutes to seventy. How forwarded records form, why rebuilds are a treadmill, and when a heap is…
We migrated from SQL Server 2014 to 2019, flipped compatibility level to 150 with the upgrade, and three month-end queries regressed forty-fold under the new cardinality estimator. The rollback that worked, the Query Store methodology that made the second attempt…
Every forty seconds, like clockwork, write throughput on our busiest primary collapsed to near zero for two to three seconds, then recovered. The culprit was not a query, a lock, or the disks. It was a redo log sized for…
A p90 latency report drifted 40ms after a rewrite for MySQL compatibility, because PostgreSQL's percentile_cont interpolates and the hand-rolled MySQL replacement picked the nearest row. Window functions look portable across the two engines; the frame, aggregate, and ordered-set corners are…
The store-locator endpoint had a spatial index on both engines and was doing a full scan on both — because ST_Distance_Sphere in a WHERE clause uses no index anywhere. Then a bulk load died at row 840,000 with ERROR 3643…
A ported importer hit one duplicate key at row 312 of a 2-million-row batch and failed the entire run — because code that coasts on MySQL's statement-level rollback meets PostgreSQL's aborted-transaction state and dies. Savepoints look identical in the manual;…
A bad data merge closed a loop in a category tree, and the two engines failed differently: MySQL died at iteration 1001 with ERROR 3636, PostgreSQL spun for 38 minutes until statement_timeout cut it. Recursive CTEs share a syntax but…
A tenant-scoped range scan over the events table read 4.1 million pages to return columns that fit in 500 bytes per row — the other 1.4 KB was a jsonb payload nobody selected, sitting inline in every heap page. Setting…