Notes for the problems that show up after launch: bad plans, awkward migrations, index debt, vacuum pressure, replica lag, and the small decisions that make PostgreSQL easier to operate.
A replica that cannot survive kill -9 is a replica you rebuild at 3 a.m. This covers relay_log_recovery, the repository history that made replicas crash-safe, GTID auto-positioning, and why you should crash replicas on purpose.
InnoDB FULLTEXT and PostgreSQL tsvector both promise search without a search engine. Here is how indexing, ranking, and language support actually compare, and when neither is enough.
Partitioning solves data-lifecycle problems, not performance problems. Field notes on verifying pruning, automating partition maintenance, living with the foreign-key restriction, and knowing when to say no.
MySQL cannot index a JSON column directly, and the workarounds have sharp edges. Field notes on generated columns, multi-valued indexes, MEMBER OF pruning, and when to stop pretending and promote the field.
Random UUIDs punish InnoDB's clustered index far harder than PostgreSQL's heap, but neither engine loves them. Here is the real damage model, and how UUIDv7 and ULID change the math.
A selective query took 1.5 seconds as a parallel seq scan on NVMe and 45 ms as an index scan — same data, same index, one GUC apart. random_page_cost 4.0 is seek-time math running on hardware with no seek time.
Both MySQL and PostgreSQL replicas serve stale data by default. What differs is the toolbox: GTID waits and group replication consistency levels versus LSN tokens and remote_apply.
Concurrent FOR SHARE locks and foreign key checks consume MultiXact IDs, and multixact has its own wraparound emergency separate from XID wraparound. How to monitor mxid_age(datminmxid), SLRU pressure, and vacuum's freezing before it stops your database.
MySQL JSON with generated-column indexes and PostgreSQL JSONB with GIN solve the same problem with different tradeoffs. Field notes on operators, updates, and indexing from someone who has run both.
full_page_writes protects you from torn pages but can dominate WAL volume. How checkpoints set the FPI rate, what wal_compression with lz4 or zstd buys you, and how to measure WAL with pg_stat_wal.
The planner is only as good as its statistics. When autoanalyze lags behind a bulk load, the bad plans arrive long before the next analyze run does.
The JSON keyword in MariaDB is a compatibility alias over LONGTEXT. It migrates schemas cleanly, then surprises you at scale: full parses per row, whole-document rewrites per update, and equality semantics that differ from MySQL.