A Look at the Elephant's Trunk PostgreSQL 13
summary
PostgreSQL 13 lists concrete, deployable changes: drop database with force actually disconnects clients, autovacuum now triggers on inserts and can run index vacuuming in parallel, and ANALYZE and backups expose progress views so long operations report percentage done. UUID generation moves into core as gen_random_uuid, logical replication finally handles partitioned tables, backups include a standard manifest and bg_verify_backup for integrity checks, and planner/runtime features like incremental sort and disk-based hash aggregates reduce memory and I/O.
AI-written summary · about
chapters (23)
- 0:00A look at the Elephants Trunk PostgreSQL 13
- 1:44Development schedule • July 2019 - branch 12
- 4:37New features • DBA and administration • SQL and developer • Backup and replication
- 5:19DROP DATABASE • Drop database • And automatically disconnect users!
- 6:47Wait events renamed • Many wait events renamed
- 8:12ALTER STATISTICS • Statistics target on extended statistics • Separate from underlying columns
- 9:22ANALYZE • Progress report for analyze • How far along is the analyze • How much has been generated
- 10:02Parallel VACUUM
- 12:59Slow query sampling
- 15:03SSL versions
- 17:14pg_stat_activity • New column leader_pid • Shows leader in parallel query • NULL for non-parallel query
- 17:58Trusted extensions • Extensions installed as non-superuser
- 19:21UUID • Random generation now in core • Covers most use-cases • No need for extension
- 19:58JSONPATH . Now supports datetime
- 20:36FETCH FIRST WITH TIES
- 22:35Backup manifests Included in base backups . By default • List of files, checksums
- 23:34Logical replication of partitioned tables • Replicates individual partitions
- 24:27Limit slot disk usage
- 25:04Online reconfiguration • primary_conninfo and primary_slot_name • Change in file or with ALTER SYSTEM . Only a reload needed
- 26:29Disk based hash aggregate • Hash aggregates used in more cases . Even when work_mem is not enough • Avoid OoMing on bad estimates • GROUP BY, DISTINCT and grouping sets
- 27:42pg_stat_statements
- 28:19Deduplication in btree • Indexes with many duplicates - Full set of columns • Improved performance
- 29:05There's always more
related talks
55mA look at the Elephants Trunk: PostgreSQL 14
50mA look at the elephant's trunk - PostgreSQL 10
52mA look at the Elephants Trunk - PostgreSQL 15
43mPostgreSQL 17 and beyond
46mPostgreSQL 16 and beyond
54mBreakout Session
19mVacuuming Enhancements in PostgreSQL 18: Faster, Smarter, More Predictable
49mSELECT 'amazing features' FROM "postgresql"