Read replicas help us to scale our read operations
We can create up to 15 read replicas within the same AZ, cross AZ or cross region
Replication is async between the RDS and replicas => reads are eventually consistent
Replicas can be promoted as standalone databases
Read Replica Use Cases
Read replicas are used for reading => only SELECT kind of statements should be executed on them
Example of use case: analytics on top of the main database which can cause performance degradation of the main DB
Read Replica Network Cost
In AWS there is a network cost when data goes from one AZ to another
Exception for managed services, since RDS read replicas are considered to be managed services, within the same region we don’t pay a free for inter-AZ data transfer
RDS Multi AZ (Disaster Recovery)
It is a synchronous replication to another stand-by database in another AZ
We have only one DNS name for the database, the failover in case of a disaster happens automatically => will increaser availability
Failover happens automatically in case of an AZ outage, loss of networking connectivity or instance storage failure
No manual intervention required
Not used for scaling!
The Read Replicas can also have Multi AZ synchronous replication enabled
From Single-AZ to Multi-AZ
In case we go for a database from single-AZ setup to multi-AZ:
There is no downtime (no need to stop the DB)
Only thing to do is to click on modify for the database
The following will happen internally:
A snapshot is taken automatically
A new DB is restored automatically from the snapshot in a new AZ
Synchronization is established between the two databases
Aurora Replicas - Auto Scaling
Clients must use the writer endpoint for accessing the writer; they must connect to the reader endpoint to access the read replicas
Global Aurora
Aurora Cross Region Read Replicas:
Useful for disaster recovery
Simple to set up
Aurora Global Database (recommended)
1 region is defined as the primary region (can be used for reads and writes)
Up to 5 secondary regions, can be used for reads only. Replication lag is less than 1 second
We can have up to 16 read replicas per secondary region
In case of a disaster, the RTO for promoting another region is less than 1 minute
Typical cross-region replication takes less then 1 second
