When zero-downtime saves lives: the reality of Active-Active Postgres
October 20–23
Level: Intermediate
What if downtime means unanswered emergency calls and lives at risk? This is the reality for a geo-redundant emergency system operating across island datacenters with spotty connectivity. While 99.9% of applications don't need Active-Active databases, this life-critical system lives in the remaining 0.1%. To survive, each site must run autonomously while remaining synchronized via Active-Active PostgreSQL.
Setting up bidirectional logical replication is easy for a demo. The challenge comes with real traffic, network partitions, and major database upgrades.
In this talk, we bridge the gap between high-stakes requirements and distributed-systems reality:
- How cross-functional teams designed a near-zero-downtime major upgrade.
- Lessons from production load and split-brain scenarios using EDB Postgres Distributed.
- Moving from conflict detection to real-world conflict resolution.
Join for a practical blueprint for deploying and upgrading Active-Active Postgres.
