PostgreSQL on ZFS
summary
ZFS provides instant snapshot pointers that let Postgres get point‑in‑time clones without copying data, snapshots incur near‑zero runtime impact, and clones become writable for testing or table restores. Incremental send/receive ships only changed disk blocks for efficient daily or 10–20 minute replication; initial clone can be slow but increments are small. Built‑in compression often improves throughput by reducing I/O, pools handle hot‑drive upgrades with zero downtime, and OmniOS/illumos/FreeBSD/Linux builds exist with an OmniOS VM and example scripts to reproduce demos.
AI-written summary · about
chapters (9)
- 0:00Start
- 2:57Cloning for Validation and Testing
- 5:59Disaster Recovery and Backups
- 6:39The Human Disasters
- 10:24How Zfs Is Structured
- 11:29Upgrading Physical Drives with Zero Downtime
- 24:14Compression
- 39:37Summary
- 41:36Deduplication
related talks
17mA Laboratory for Databases
46mPostgreSQL 16 and beyond
25mVindicating ZFS with PostgreSQL: Unleashing the Power of Scalability
35mVindicating ZFS with PostgreSQL: Unleashing the Power of Scalability
42mPostgreSQL Distributed: Architectures & Best practices
55mHigh Availability with PostgreSQL and Pacemaker
38mPerforming Databases Migrations Without Overtime
14mBuilding a better PostgreSQL together