PGCon2016 - 20180510
PGCon 2016
The PostgreSQL Conference
| Speakers | |
|---|---|
|
Josh Berkus |
| Schedule | |
|---|---|
| Day | Talks - Day 2 - 2016-05-20 |
| Room | DMS 1120 |
| Start time | 13:00 |
| Duration | 00:45 |
| Info | |
| ID | 973 |
| Event type | Lecture |
| Track | Scaling Out |
| Language used for presentation | English |
How I Learned to Stop Worrying and Love Failover
Automated HA on a container stack
Have you been told that automated failover for PostgreSQL is "too hard" or even "impossible"? Have you been forced to live with SLAs measured in hours? Is spinning up a new replica something which requires weeks of planning? Well, no more. Using Project Atomic, Kubernetes and Patroni, you can have always-up PostgreSQL-as-a-Service in your own datacenter. And surprisingly easily, too!
Thanks to many of the new tools surrounding Linux containers, HA PostgreSQL clusters are now easier than ever before. This presentation will demonstrate an autofailover cluster of PostgreSQL-as-a-Service, and then detail components, configuration, and usage. Our autofailover solution is built with:
- Project Atomic
- Kubernetes
- Patroni
- Etcd
Don't listen to the naysayers! PostgreSQL HA is achievable with low risk and reasonable effort. I'll explain how.