When the Main Database Goes Down
Promote a replica, fail over, and learn which writes you can lose
What you'll learn
- Promoting a replica and pointing the app at it
- How automatic failover in managed databases works
- RPO, RTO, and why recent writes can be lost
- Split brain, fencing, and testing backups
In this video, we'll look at what to do when your main database server dies and every write starts failing. We'll cover promoting a replica to be the new primary, automatic failover in managed databases, and why the last few writes before a crash can be lost.
We'll also look at RPO and RTO, how two servers can both think they're in charge (split brain) and how fencing stops it, and restoring from backups when there's no replica.
Finally, we'll look at why a backup you've never restored is only a guess, and why you should test yours.
