KubeCon Europe 2016

Why pay for always-on relational database service when you can deploy it yourself so easily? This demo-heavy talk will show off a deceptively simple high availability stack for PostgreSQL, using Docker, Etcd, Kubernetes, Patroni and Atomic. Not only is this open source solution ready to go to give you HA Postgres right now, it represents an approach which can be adapted to other relational databases with replication.

Outline:

1. Introduction
2. First demo
2.1 spin-up
2.2 connecting an application
2.3 automated failover
2.4 redeployment
3. History
3.1 Handyrep
3.2 Zalando
3.3 Compose.io
3.4 AtomicDB
4. Design
4.2 the Patroni agent
4.2 use of Etcd/Zookeeper
4.3 Kubernetes service definitions
4.4 How failover works
5. 2nd Demo
5.1 Code and configuration tour
5.2 Repeat of failover demo
5.3 Requests from audience
6. Conclusion