Elephant Herd as a Service: Managing Hundreds of Postgres Instances

Date: 2016-09-15
Time: 18:30 - 19:20
Room: Ft. Worth 1
Level: Intermediate

Replicas, and failovers, and environments, Oh My! It's all too easy for one Postgres instance to become two, and then three, and then... Then the targeted applications with their special-purpose instances on an array of ports start to pile up. Then development, QA, staging, acceptance testing, and production need their own little rules. Suddenly there are dozens or even hundreds of Postgres elephants roaming our Savannah, each in their own little specialized herd. Managing all of that manually is a massive headache that never seems to end.

Enter ElepHaaS: a Django tool meant to be a command center, medic, and organizer for all of our numerous Postgres herds. From simple starting and stopping to full disaster recovery management with DNS definition capabilities, to leader and follower cloning, and more. We built it to wrangle and expand our multitude of elephants before they trampled us. Let's talk about why we built it, how it works, what it does, and how you can use it too. It's still pretty young, but it's Open Source and free to customize or extend.

Maybe it'll even take on a life of its own, and becomes something we never even imagined. It all starts here...


Shaun Thomas