Inside PostgreSQL High Availability: Quorum, Split-Brain, and Failover at Scale
Presented by:
Venkat Akhil
Venkat Akhil is a PostgreSQL SME with extensive experience in database architecture, high availability, and security. With deep expertise in multiple PostgreSQL versions including 18, Akhil has designed and implemented large-scale high availability clusters, managed complex replication and failover scenarios using Patroni, and led enterprise-grade database migration and upgrade projects.
Passionate about PostgreSQL internals and optimization, Akhil actively explores new features, including PostgreSQL 18, conducts hands-on testing, and shares insights through technical blogging and community engagement.
In addition to technical leadership roles in the banking and finance sector, Akhil contributes to the PostgreSQL open-source community and organizes local meetups focused on PostgreSQL.
No video of the event yet, sorry!
Abstract
PostgreSQL High Availability (HA) isn’t just about replication,it’s about making distributed systems converge under failure. This session goes deep into the internals of PostgreSQL’s HA stack, focusing on how Patroni, etcd/Consul, and HAProxy collaborate to maintain consensus and avoid split-brain scenarios. We’ll dissect the timing and sequence of failover events, WAL divergence detection, and fencing techniques using watchdogs.
Through a series of simulated failures—network partitions, disk stalls, and delayed replicas,one’ll learn how PostgreSQL behaves under extreme conditions, and how to tune parameters like synchronous_standby_names, primary_conninfo, and maximum_lag_on_failover for deterministic recovery.
By the end of this talk, one’ll understand the fine line between consistency and availability, and how to engineer PostgreSQL clusters that survive chaos without human intervention.
- Date:
- Duration:
- 45 min
- Room:
- Conference:
- PGConf India, 2026
- Language:
- Track:
- Database Administration
- Difficulty:
- Medium