Reliability features in Postgres Pro
summary
PostgresPro ships an integrated HA layer called beha that uses raft consensus and a custom BCP protocol with a beha_worker process to perform automatic failovers, manual switchovers and a 2+1 referee topology; it requires no external DCS, stores cluster state in PG and elects the node with the newest LSN as leader. Backup is handled by PG Pro Backup (integrity checks, compressed incremental, parallel jobs, remote mode without SSH, a PG Pro Backup catchup mode and a FUSE mount), and libpq/HAProxy callbacks are used for client routing; deployments report hundreds–thousands TPS and successful automatic failovers.
AI-written summary · about
chapters (20)
- 0:00Introduction and agenda
- 1:56Postgres Pro reliability features
- 2:57Physical replication overview
- 5:49Built-in HA solution BHA
- 11:30BHA architecture and setup
- 16:12BHA cluster components
- 19:07Raft protocol and Quorum
- 23:50Manual switch over in BHA
- 25:58Automatic failover in BHA
- 27:52BHA benefits and summary
- 28:13Client connection to BHA cluster
- 30:27LIBPQ connection handling
- 30:47HAProxy for load balancing
- 33:41Custom solutions for connections
- 36:55Backup tools for BHA cluster
- 38:03PG Pro Backup features
- 43:12PG Pro Backup new version
- 44:59Backing up BHA cluster
- 47:29Key takeaways and conclusion
- 48:34Q&A session
related talks
47mPostgreSQL High Availability: The Considerations and the Candidates
46mDeploying Maximum HA Architecture With Postgres
50mEvolution Of PostgreSQL Replication and High Availability (upto ~ 99.999%)
36mDesigning Layered High‑Availability Architectures for PostgreSQL on a Budget
55mHigh Availability with PostgreSQL and Pacemaker
29mBuilding Enterprise Grade PostgreSQL using Percona distribution for PostgreSQL
50mThe Immortal Postgres Web
50mPatroni Deployment Patterns