Azure Site Recovery: disaster recovery of VMware virtual machines

The solution Azure Site Recovery (ASR) protects virtual or physical systems, hosted both Hyper-V environment that VMware, automating the replication process to a secondary data center or to Microsoft Azure. With a single solution you can implement Disaster Recovery plans for heterogeneous environments orchestrating the replication process and actions needed for the successful recovery. Thanks to this solution, the DR plan will be easily available in any eventuality, even the most remote, to ensure business continuity. Recently, the solution has been expanded while also providing the ability to implement a disaster recovery strategy for Azure virtual machines, allowing you to enable replication between different regions.

In this article I'll show you how ASR can be used to replicate virtual machines in VMware environment to Azure (scenario 6 in the following figure), examining the characteristics and technical procedure to be followed. The following illustration shows all the scenarios currently covered by ASR solution:

Figure 1 – Scenarios covered by Azure Site Recovery

The replication scenario of VMware virtual machines to Azure requires the presence of the following architecture:

Figure 2 - Architecture in the replication scenario VMware to Azure

In order to activate the replication process is required the presence of at least one on-premises server on which you install the following roles:

  • Configuration Server: coordinates communications between the on-premises world and Azure, and manages the data replication.
  • Process Server: This role is installed by default with the Configuration Server, but may be provided more Process Server based on the volume of data to be replicated. It acts as a replication gateway, then receives replication data, performs an optimization through caching and compression mechanisms, provides encryption and sends them to the storage in the Azure environment. This role is also responsible to make the discovery of virtual machines on VMware systems.
  • Master target server: even this role is installed by default with the Configuration Server, but for deployment with a large number of systems can be more servers with this role. Take action during the failback process of resources from Azure by managing replication data.

On all virtual machines subject to the replication process is required the presence of Mobility Service, that is installed by Process Server. It is a special agent in charge of replicating the data in the virtual machine.

Following describes the process to follow to make the deployment of on-premises and Azure components required to enable replication of VMware virtual machines to Microsoft's public cloud.

The core component required on Azure side is the Recovery Service Vault within which, in the section Site Recovery, you can start the configuration process controlled by the chosen scenario.

Figure 3 – Choice of replication scenario of VMware virtual machines within the Recovery Service Vault

Then you must install on the on-premises machine the Configuration Server by following the steps listed:

Figure 4 – Steps to follow to add the Configuration Server

In this section of the Azure portal it is possible to download the Microsoft Azure Site Recovery Unified Setup and the key required for the registration of the server to the vault. Before starting the installation make sure that the machine on which you intend to install the Configuration Server be able to access the public URLs of the Azure service and that is enabled during the setup the web traffic on port 80 needed to download the MySQL component used by the solution.

The setup prompts you for the following information:

Figure 5 – Choice of roles to install

Select the first option for installing the roles Configuration Server and Process Server. The second option is useful if you need to install additional Process Server to enable a scale out deployment.

Figure 6 - Accept the license agreement by MySQL Community Server

Figure 7 - Key selection required for the registration to the Site Recovery Vault

Figure 8 - Choice of the methodology to access the Azure Services (direct or via proxy)

Figure 9 – Check to verify prerequisites

Figure 10 – Setting passwords for MySQL

Figure 11 – Further check on the presence of the required components to protect VMware VMs

Figure 12 – Choice of the installation path

Installation requires approximately 5 GB of available space, but are recommend at least 600 GB for the cache.

Figure 13 — Select the network interface and the port to use for replication traffic

Figure 14 – Summary of installation choices

Figure 15 - Setup of the different roles and components successfully completed

At the end, the setup shows the connection passphrase which is used by the Configuration Server, that is good to save with care.

Then you must configure the credentials that will be used by Azure Site Recovery to discover virtual machines in the VMware environment and for the installation of the Mobility Service on virtual machines.

Figure 16 - Definition of the credentials used by the service

After complete these steps you can select the Configuration Server from the Azure portal and then define VMware system data (vcenter or vSphere) with which to interface.

Figure 17 - Select the Configuration Server and add vCenter / vSphere host

On completion of this configuration it is necessary to wait few minutes to allow the Process Server to perform the discovery of VMware virtual machine on the specified environment.

Then you need to define the settings for the target of the replica:

  • On which subscription and what recovery model (ASM or ARM).
  • Which storage account use to host the replicated data.
  • vNet on which attest the replicated systems.

Figure 18 – Target replication settings

The next step involves defining the replication policy in terms of RPO (in minutes), retention of the recovery points (expressed in hours) and how often make consistent snapshot at the application level.

Figure 19 – Creation of the replication policy

Upon completion of this task is proposed to carry out the analysis of your environment using the tool Deployment Planner (available directly through the link in the Azure Portal) in order to ensure that the requirements, network resources and storage resources are sufficient to ensure the proper operation of the solution.

Figure 20 - Steps of infrastructure preparation completed successfully

After completing the infrastructure preparation steps you can activate the replication process:

Figure 21 - Source and Replica Target

Figure 22 - Selection of the virtual machines and of the related discs to be replicated

This section also specifies which account the Process Server will use to install the Mobility Service on each VMware virtual machine (account configured previously as documented in Figure 16).

Figure 23 - Replication policies selection and optionally enable Multi-VM consistency

If the "Multi-VM consistency" option will be selected it will create a Replication Group within which will be included the VMs that you want to replicate together for using shared recovery point. This option is recommended only when you need a consistency during the fail over to multiple virtual machines that deliver the same workload. Furthermore, by activating this option you should keep in mind that to activate the system failover process is necessary to set up a specific Recovery Plan and you can not enable failover for a single virtual machine.

At the end of these configurations you can activate the replication process

Figure 24 – Activation of the replication process and its result

Figure 25 - State of the replica for the VMware virtual machine

One of the biggest challenges when implementing a Disaster Recovery scenario is to have a chance to test its functionality without impacting production systems and its replication process. Equally true is that do not test properly the DR process is almost equivalent to not having it. Azure Site Recovery allow you to tests in a very simple way the Disaster Recovery procedure to assess the effectiveness:

Figure 26 – Testing the Failover procedure

Figure 27 - Outcome of the Test Failover process


Being able to rely on a single solution as Azure Site Recovery that lets you enable and test procedures for business continuity in heterogeneous infrastructures, contemplating even virtual machines in VMware environment, certainly has many advantages in terms of flexibility and effectiveness. ASR makes it possible to deal with the typical obstacles encountered during the implementation of Disaster Recovery plans reducing the cost and complexity and increasing the levels of compliance. The same solution can also be used to deal with the actual migration to Azure with minimal impact on end users thanks to nearly zero application downtime.