Q: When you perform a failback from Azure to on-premises via Azure Site Recovery, how is data sent back?
A: When an operating system fails over to Azure, reverse replication isn't activated as an ongoing activity from Azure back to on-premises. This is because once the virtual machine is running in Azure, it's stored in georeplicated storage—which means the virtual machine already has resiliency built in while running in Azure. When a failback to on-premises from Azure is required, the failback process checks which blocks have changed since the initial failover to Azure and starts sending those changed blocks back to on-premises, while the virtual machine continues to run in Azure. Once the changes are mostly copied back to on-premises, the virtual machine is stopped in Azure, the remaining changed blocks are sent to on-premises, the virtual machine is failed back to on-premises, and the replication from on-premises to Azure is then activated again. Sending the blocks from Azure to on-premises while the virtual machine is running minimizes downtime.
