We have recently posted an article explaining the Backup and DR for Azure Disks. We encourage Azure IaaS users to refer to this document for planning the right Backup and Disaster Recovery (DR) methodology for their disks. Following are a few considerations.
Durability of Azure Disks
We are excited to report that Azure has consistently delivered enterprise-grade durability for IaaS disks, with an industry-leading ZERO % Annualized Failure Rate. This means that Azure has never had a permanent failure of an IaaS disk. For Azure customers, it directly translates into lower cost, better reliability, and smoother operations for the critical applications running on Azure. For example, you don’t have to build costly RAID solutions with redundancy in most cases because Azure protects your data with three redundant copies stored locally. Even if two different hardware components that hold your disks fail at the same time, durability of your data is protected and Azure automatically spawns new replicas in the background to replace the lost copies. We built the Azure disks platform with the core consideration that protecting the data is critical to any storage platform. With Azure disks, you won’t have to constantly worry about losing your disks in the cloud.
Hardware faults can sometimes result in temporary unavailability of the VM, which is covered by Azure SLA for VM availability. Azure also provides an industry-leading SLA for single VM instances that use Premium Storage disks.
Backup and Disaster Recovery
While the platform has built-in protection for localized hardware failures, you must still plan for Backup and Disaster Recovery for safeguarding against major incidents which can cause large-scale outages. This includes catastrophic events like hurricane, earthquake, fire etc. For DR, you should plan the periodic backup of your data to a different geographic location. Please refer to Backup and DR for Azure Disks article for details.
An important thing to consider for Disk backups is to take “consistent backups”. The backups must be taken at a coordinated, consistent state of all the disks on a VM. When possible, this must be coordinated with the applications also to produce “application consistent” backups. This is necessary to make sure you can restore the VM and the application to a valid state at the time of recovery.
Azure Backup Service can be used as the Backup solution for your disks, and it works with Managed Disks as well as Unmanaged Disks. Backup Service handles the coordination of disks for consistent backup and offers GRS option for the vault which replicates the backup to a different geographic region for DR.
Another solution is to create “consistent snapshots” for the disks. In this case, you have to handle the coordination of disks for creating consistent backups, and the replication of backups to different geographic location.