Migrate MySQL on-premises to Azure Database for MySQL: Business Continuity and Disaster Recovery (BCDR)
APPLIES TO: 
Azure Database for MySQL - Single Server
Azure Database for MySQL - Flexible Server
Prerequisites
Back up and restore
As with any mission critical system, having a backup and restore and a disaster recovery (BCDR) strategy is an important part of your overall system design. If an unforeseen event occurs, you should have the ability to restore your data to a point in time (Recovery Point Objective) in a reasonable amount of time (Recovery Time Objective).
Backup
Azure Database for MySQL supports automatic backups for 7 days by default. It might be appropriate to modify this to the current maximum of 35 days. It's important to be aware that if the value is changed to 35 days, there are charges for any extra backup storage over 1x of the storage allocated.
There are several current limitations to the database backup feature as described in the Backup and restore in Azure Database for MySQL docs article. It's important to understand them when deciding what additional strategies that should be implemented.
Some items to be aware of include:
No direct access to the backups
Tiers that allow up to 4 TB have a full backup once per week, differential twice a day, and logs every five minutes
Tiers that allow up to 16 TB have backups that are snapshot-based
Note
Some regions do not yet support storage up to 16TB.
Restore
Redundancy (local or geo) must be configured during server creation. However, a geo-restore can be performed and allows the modification of these options during the restore process. Performing a restore operation can temporarily stop connectivity and any applications is down during the restore process.
During a database restore, any supporting items outside of the database need to be restored.
Review the migration process. See Perform post-restore tasks for more information.
Read replicas
Read replicas can be used to increase the MySQL read throughput, improve performance for regional users and to implement disaster recovery. When creating one or more read replicas, be aware that additional charges are applied for the same compute and storage as the primary server.
Deleted servers
If an administrator or bad actor deletes the server in the Azure portal or via automated methods, all backups and read replicas are deleted. it's important that resource locks are created on the Azure Database for MySQL resource group to add an extra layer of deletion prevention to the instances.
Regional failure
Although rare, if a regional failure occurs geo-redundant backups or a read replica can be used to get the data workloads running again. it's best to have both geo-replication and a read replica available for the best protection against unexpected regional failures.
Note
Changing the database server region also means the endpoint does change and application configurations needs to be updated accordingly.
Load balancers
If the application is made up of many different instances around the world, it might not be feasible to update all of the clients. Utilize an Azure Load Balancer or Application Gateway to implement a seamless failover functionality. Although helpful and time-saving, these tools aren't required for regional failover capability.
WWI scenario
WWI wanted to test the failover capabilities of read replicas so they performed the steps outlined below.
Create a read replica
Open the Azure portal.
Browse to the Azure Database for MySQL instance.
Under Settings, select Replication.
Select Add Replica.
Type a server name.
Select the region.
Select OK, wait for the instance to deploy. Depending on the size of the main instance, it could take some time to replicate.
Note
Each replica incurs additional charges equal to the main instance.
Fail over to read replica
Once a read replica has been created and has completed the replication process, it can be used for failed over. Replication stops during a failover and makes the read replica its own main instance.
Failover Steps:
Open the Azure portal.
Browse to the Azure Database for MySQL instance.
Under Settings, select Replication.
Select one of the read replicas.
Select Stop Replication. This breaks the read replica.
Modify all applications connection strings to point to the new main instance.
BCDR checklist
Modify backup frequency to meet requirements.
Setup read replicas for read intensive workloads and regional failover.
Create resource locks on resource groups.
Implement a load-balancing strategy for applications for quick failover.