Q: What is Azure Site Recovery?

A: Azure Site Recovery is a cloud service that can be licensed on a per-virtual machine (VM) or per-physical instance basis. It enables the replication and orchestration of failover processes between a company's own physical sites to Azure services. Azure Site Recovery provides several "channels" that give organizations the ability to enable disaster recovery orchestration and replication regardless of their infrastructure. The channels that are currently available include the following:

  • Hyper-V Replica (HRM) site-to-site. This channel is the original service available. It provides an Azure portal-driven initial configuration of replication using Hyper-V Replica between a customer's own locations and then an orchestrated failover. This service works by communicating with on-premises System Center Virtual Machine Manager instances that are managing the Hyper-V hosts. HRM performs the configuration and failovers via System Center Virtual Machine Manager. The use of System Center Virtual Machine Manager also enables other types of actions to be performed by calling scripts stored in the System Center Virtual Machine Manager library. Using Hyper-V Replica, any workload running in a Hyper-V VM can be replicated because the replication is completely transparent to the guest OS. Hyper-V Replica provides additional benefits such as periodic application-consistent snapshots through integration with VSS and can inject alternate IP configuration to the guest OS in the event of a failover to match the disaster recovery data center's IP scheme.
  • Hyper-V Replica (HRM) site-to-Azure. This channel builds on the original HRM feature set for site-to-site but enables the replica VM to be hosted in Azure IaaS. The Hyper-V Replica technology is still used, providing a guest OS transparent replication and enabling replication of any guest OS that's supported in Azure. The source Hyper-V VMs must be Generation 1 and must be compatible with Azure IaaS. Like the on-premises Hyper-V Replica technology, the changes to the virtual hard disks of the VM are replicated to Azure at a defined interval, and the option to have periodic application-consistent snapshots is available. As part of the configuration of the Azure-targeted HRM, networks in the Azure Virtual Network are selected to map to on-premises virtual networks, enabling VMs in Azure to get the required IP connectivity. The actual VM isn't created in Azure until a failover is performed. When created during a failover, the VM automatically gets an IP address for its new virtual NIC from whatever virtual subnet it has been mapped to.
  • SQL AlwaysOn. This is an application-level replication technology offering asynchronous and synchronous replication between SQL Server machines. Although it's part of SQL Server, it's considered to be another channel, to be used as part of Azure Site Recovery. When possible, it's typically preferable to use application-aware replication; however, other considerations exist, such as the fact that SQL AlwaysOn requires the replication target to be running. In Azure, this means paying for the VM—whereas technologies such as HRM to Azure don't require the target VM to be running.
  • InMage Scout VMware-to-VMware. InMage Scout is a recent Microsoft aquisition. This in-OS replication technology is the industry leader in its space. InMage Scout uses an agent in the OS being replicated to fracture off writes to disk before they happen and then send those writes to the InMage Scout Process Server, resulting in almost no additional overhead on the replicated system while enabling near-synchronous replication. The InMage Scout Process Server does the heavy lifting of any caching, compression, and encryption and then sends the data to its corresponding InMage Scout Master Target Server at the target location, which then performs the writes to targets and enables various point-in-time retention points to be enabled. InMage Scout supports Windows and Linux OSs. This enables the replication of VMs running on VMware to a VM running on another VMware server. InMage Scout is available through Azure Site Recovery when specifying the VMware-to-VMware scenario. The InMage servers can be physical or virtual and run on Windows Server 2012. Multi-tier application consistency is available through protection plans.
  • InMage Scout Anything-to-Azure/Anything-to-Hyper-V. Building on the InMage Scout in-OS replication, another use case is running the agent in the OS and replicating to a VM in Azure. Because InMage Scout runs within the OS being replicated, it's completely agnostic of the platform hosting the OS instance—which means the OS could be on a VM running VMware or XenServer, the OS could be running on bare metal and not virtualized, or the OS could be running in another cloud service such as Amazon Web Services. If the OS is supported by InMage Scout, then that OS can be replicated to Azure or Hyper-V. Although the focus of Azure Site Recovery is disaster recovery, it's easy to see that the same technology could also be used as part of a migration effort.

What this essentially means is that Azure Site Recovery now enables disaster recovery for any environment, while ensuring that organizations can use the replication channel that's the best fit based on the workload and platform, and allowing a single Azure-based cloud service to facilitate the configuration and failover processes. You can perform replication from Hyper-V to Hyper-V, Hyper-V to Azure, VMware to VMware, VMware to Azure, and just about any other combination you can think of with Azure or Hyper-V as the target. The following figure summarizes Azure Site Recovery's replication and disaster recovery capabilities.