Countless caveats, cautions and priorities can guide workload protections, but consider incorporating some VM backup best practices to help improve your processes.
Use caution with backup agents
Although modern backup software can recognize and accommodate virtualized environments, experts debate the role of backing up VMs as though they were physical machines. Instead, many emphasize the use of virtualization layer backups of entire VM files, such as VMDK files. It's worth evaluating deployment performance to ensure any backups implemented at the OS level minimize the performance load on the application and host system. This is particularly crucial when multiple VMs use the same physical host, and backing up multiple VMs simultaneously could congest network communication.
Avoid snapshots for long-term backups
Snapshots are a quick and time-efficient approach for protecting many common VMs over the short term, enabling the hypervisor to use a differencing disk file to record changes rather than the original disk. Snapshot technology enables you to easily restore previous VM states, but this technology also uses large amounts of additional storage to track changes. Any storage problems can affect the integrity of a snapshot or restoration. You should limit the number of snapshots for each application and regularly consolidate snapshots into the original VM disk file.
Use incremental backup technologies
Incremental backup technologies, such as Changed Block Tracking (CBT), can vastly accelerate backup speeds. CBT tracks any storage blocks that change between backup processes, enabling the backup application to query the hypervisor -- such as VMkernel -- to determine which blocks have changed and then back up only those changes. This is far faster and easier than relying on the backup application to track changes.
Quiesce vital applications
Transactional applications, such as email servers and databases, typically must be quiesced -- paused -- before you back them up. This enables the creation of application-consistent backups that ensure the system writes all pending writes to the disk before you perform the backup. Such quiesced backups also ensure that no data is lost if you must restore the backup. However, such quiescing behavior requires that the backup tool, virtualization drivers and application all work together to observe the proper pause-and-resume behavior. You must test and verify this behavior to ensure that everything works as expected.
Backups are only useful when accessible and secure. Select the backup target location with care. As a rule, the target should be a physically remote location from the production system. Otherwise, the same event that compromises the production system -- such as flood or fire -- could also compromise the backup. In the early days of tape backups, businesses frequently transported physical tapes to an off-site location or even shipped them to a remote storage facility. Nowadays, remote replication and cloud-based storage protect backup accessibility. You can also employ encryption to secure backups against theft or loss, preventing valuable data from exploitation and possibly exposing the organization to costly regulatory compliance breaches.
Test restoration regularly
VM backup best practices don't guarantee successful restorations, so test recovery processes on a regular basis. In most cases, you can restore VM backups to test servers, where you can verify the recovery process and VM integrity without affecting the production environment. Testing not only validates the backup, but also reinforces the recovery process -- the workflow -- and strengthens the skill sets of IT staff.
Now that you're familiar with these VM backup best practices, discover what tools are available to help you. You can also explore the various backup methods for your VMs or some command-line tips for backing up your ESXi host.
Dig Deeper on Backing up VMware host servers and guest OSes
Related Q&A from Stephen J. Bigelow
Containers have rapidly come into focus as a popular option for deploying applications, but they have limitations and are fundamentally different ... Continue Reading
ALM and SDLC both cover much of the same ground, such as development, testing and deployment. Where these lifecycle concepts differ is the scope of ... Continue Reading
Eliciting performance requirements from business end users necessitates a clearly defined scope and the right set of questions. Expert Mary Gorman ... Continue Reading