Sergey Nivens - Fotolia
VMware VM backup strategies combine tradition with new methods
Different VM backup and restoration approaches can favor consistency, speed or simplicity. Consider which you value most, and tailor your methodology to that.
When it comes to a VMware VM backup, organizations can adopt several approaches. Businesses can choose to back up VMs like physical machines or as VM files -- or to tailor a combination of these approaches to each workload.
Traditional VM backups
First, IT administrators can back up and restore VMs like physical machines. This approach works because each VM includes its own OS that exists in complete isolation from other VMs. In most cases, admins can accomplish traditional backups by installing a suitable backup agent in the desired VM. The backup agent then moves VM files and data across the network to available backup servers.
One notable benefit of this traditional approach is consistency; an organization can basically use the same approach for physical machines and VMs in the data center. The backup software itself must be capable of identifying and working with VMs, but admins can often use it to quiesce applications during the backup process, enable both incremental and differential backups, and support file-level backups and restoration. However, when using this physical approach, the system might protect the OS and application components but not the VM as a complete entity. In addition, traditional backups can place significant loads on the system and network.
Protect workloads as VM files
The second approach to a VMware VM backup is to protect workloads as VM files, such as VMDK files. Such backups take full advantage of the virtualization environment, backing up or restoring entire VMs as a single, quick and cohesive step. In most cases, this approach relies on snapshots to periodically capture a VM's state and then uses the change log files to commit changes to the VMDK file later.
The main benefit with this approach is speed and simplicity. An admin can capture a VM file in almost real time so it doesn't affect system overhead or network traffic. Additionally, when VM files use shared storage, such as a storage area network or network-attached storage, an admin can employ replication-type protection to back up VM files without ever placing additional load on the system running the VM.
However, capturing the VM file captures everything in it: all of the OS components, application files and working data. Recovery means restoring the entire VM file because there's no means of recovering individual file components. Consequently, restorations process more data -- because VM files can be large -- and consume more time than recovering an individual file using more traditional approaches.
So, which approach is best for a VMware VM backup? Both have advantages and disadvantages, so an admin must consider the tradeoffs involved, as well as the importance and recovery point objective/recovery time objective needs of each workload, before adopting either method. These approaches aren't mutually exclusive, so an admin could employ both in the same environment to best suit the needs of each workload.