I believe, In case of Multi-AZ RDS the commit happens ( or confirmed) after it has been replicated to slave copy ? Write to master and Replicate to redundant DB copy and then commit successful ?
3 Answers
I think a commit is only successful if data is written to both primary and standby database instances.
Per this blog post linked below, the commits with Multi-AZ incur a latency is between 2ms to 5ms so it’s likely not long enough to impact normal workloads.
https://aws.amazon.com/blogs/database/amazon-rds-under-the-hood-multi-az/
@AK it may depend which database do you have in mind (RDS for PGSQL? Aurora?). As far I know the RDS for PostgreSQL support "synchronous_commit", so the master waits for one or more replicas to confirm the transaction. https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/AuroraPostgreSQL.Reference.html In the course it’s said "Replication between masters is synchronous, to read-replicas is asynchronous"