I'm implementing disaster recovery plans on my database and I have a few concerns with how Read Replicas work fundamentally.

Let's say I have a bunch of order information but then the database crashes, from my understanding I could essentially just swap out the replica for the source database and I will have all the database information and can continue.

The scenarios that I am unclear about are that let's say someone accidentally(or intentionally) deletes a bunch of data and I want to fall back to the read replica, if the data is deleted on the source database will it get deleted on the read replica as well?

Essentially I want to know exactly what data is going into the read replica and if I can restore it to the exact point in time I need

  • This is more dependent on the specific RDS engine and type of replica than you might expect. Which flavor of RDS are we talking about? – Michael - sqlbot Mar 23 at 21:36

Read replicas do not help with preventing corruption. They're used to provide a "read replica" to allow you to direct database read traffic to one or more asynchronous copies of the main database. All writes must go to the master. You must configure your application(s) manually to use read replicas.

RDS Multi-AZ deployments are used to increase reliability - if there's a problem with the primary database DNS is changed to point to the standby, which is promoted to master, and a new replica is created. There's a period of interruption of 1-5 minutes, but there's typically no data loss.

Your second question is about restoring deleted data. That's what backups are for. RDS provides backups and can do point in time restores.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.