In my previous article, I have spoken about the possibilities of leveraging multiple cloud providers at the same time, for the sake of high-availability and disaster recovery. Today I want to take a look at some details on how the technology works.
In my scenario, I have two simple virtual machines. The one called “Webserver” actually contains a freshly installed web server (IIS). The second (management server) will be required for the migration setup.
For the purpose of this test, I will need a storage account for Azure Recovery Services. This storage account will store the replicated disks.
In order to use the Recovery Services, you will need a recovery vault. In the new portal, you add a new resource of the Backup and Site Recovery (OMS) category:
Creating the Vault
The wizard will walk you through the whole process:
The connection between AWS and Azure will be achieved through the Configuration Server, which required the above mentioned registration key. The steps to setup the software are really simple.
Once you have it set up, you may want to add accounts to the configuration server so you can deploy the mobility agents to the VMs from a central location. Otherwise, you may deploy them manually.
The last part the of setup is to create replication policies:
Before you start replication of the machine, it is important to add the Azure VM Agent, just so you can manage the machine through the Azure backend (using extensions, for example):
From this point forward, it is about enabling replication for the VMs:
Since they are considered ‘Physical machines’ for all intents and purposes, you will need to manually point to each one of them:
Once the job finishes:
You will still need to wait for the initial replication to finish.
Once your VM is protected, you can try a test failover, which will create a machine in Azure, completely independent from the original one, in a specific network (for isolation):
Since this new machine doesn’t have any external access method, I have used a PowerShell script to and from the required endpoints (RDP if needed and Port 80):
Once done, you can see the website running in the test machine.
After this, the steps will involve:
- Protecting all required VMs
- Designing a recovery plan that considers all aspects:
- AD requirements
- External IPs and External DNS
As you can see, the steps are pretty straightforward. The bulk of you work will be designing the proper plans, testing and documenting.