Problem solve Get help with specific problems with your technologies, process and projects.

How to handle the server backup process in the virtual, physical world

Virtualized backup tools sport unique features but lack the functionality necessary to handle the physical server backup process.

Since many companies still haven't virtualized 100% of their servers, the server backup process has to keep up with a mixed physical and virtual environment.

The momentum propelling virtualization deeper into the data center is strong. Statistics vary, but none of the survey data yet shows that the majority of companies have switched to all-virtual data centers.

A data center with a combination of physical and virtual servers is the reality for many of today's IT teams.

Many organizations still have some servers running on traditional physical hosts and some running as virtual machines (VMs). While there are benefits to virtualizing all workloads, this mix of the physical and the virtual is the reality IT admins must deal with today. When it comes to protecting this mixed environment using traditional backup processes, data center administrators are left working with an awkward combination of backup tools.

Traditional and virtual server backup basics

With traditional physical servers, backup is accomplished through these tactics:

  • File-based backup: This is where traditional agent-based tools perform backups of the files on the guest operating system. The method is easy to understand. The downside is that files can be opened and locked, resulting in incomplete backups. You also must install an agent on each server being backed up.
  • Full backup/archival backup: The archive bit is a file attribute in the Windows OS that indicates whether or not a file has been backed up. Typically the terms full backup and archival are used synonymously to indicate a full, file-based backup of all files on a server.
  • Incremental backup: This backup method saves only the files unchanged since the last incremental backup. Thus, if daily backups were done, a restore would include the archival plus the daily incrementals of changed files.
  • Differential backup: This is a file-based backup of all files changed since the last archival. Differentials grow each day, but when it comes time to recover, you'll only need to recover the archival plus the last differential.

With virtualization, backup involves the following concepts:

  • Image-level backup: This is a full backup of a virtual machine's virtual disk file. It could be created by simply copying the virtual disk or by combining changed blocks of the virtual machine's disk file with the previous full backup.
  • Synthetic backup: This involves the combination of changed blocks (similar in concept to a file incremental) with the original full backup of the virtual machine disk. It creates the most recent and restorable image-level backup of a virtual machine.

Physical vs. virtual server backups

A data center with a combination of physical and virtual servers is the reality for many of today's IT teams. Here are key differences between the server backup process for each.

Physical server backups: A traditional physical server backup works on the concept of installing a backup agent on each host, performing an initial full backup over the network and then on subsequent nights backing up just the files that were changed. These agents typically offer options to back up specific applications, such as Windows Active Directory, SharePoint or Exchange.

Traditional physical server backups work at the file level, so the restoring is done based on the file or directory you want to restore -- unless you have a special application agent that might, for instance, restore just a single Exchange Server mailbox.

While you can use a traditional method to back up your VM, such as installing an agent inside the operating system, that can create bottlenecks. The different architecture of the virtual infrastructure can cause performance problems when you attempt to back it up with traditional methods that have no visibility into or understanding of the virtual infrastructure.

Virtual server backups: Virtualization-specific backup, on the other hand, works much more efficiently because it typically operates outside the server operating system (without installing an agent).

Virtual server backup tools typically use a vendor's application programming interface to obtain just the changed blocks of the virtual machine disk file and back up only those blocks, and it can be done during the day without creating downtime for users. Additionally, those blocks could be replicated to a secondary off-site data center, fulfilling your need to store backup data elsewhere.

While virtualization backup tools normally don't use agents, they still offer varying levels of Microsoft's volume shadow copy services (VSS) to quiesce Windows virtual machines and applications. That ensures that business-critical applications inside those virtual machines have integrity and work as expected.

Backups of virtual machines using virtualization backup software can be restored at either the image level (the entire VM disk) or file level (individual files from the VM's virtual disk).

Another benefit of virtualization-specific backups is that they talk to the centralized virtual infrastructure management application, such as VMware's vCenter, to discover what virtual machines are out there. They also find out what hosts those machines are running on, since it's likely the VMs are moving from host to host using a feature such as VMWare DRS (Distributed Resource Scheduler).

With a virtualization backup tool, you can choose to back up all virtual machines at a certain hierarchical tree of the virtual infrastructure (like the virtual data center level or resource pool level or cluster level). Once server backup processes are enabled at a specific level, backups of all virtual machines at that level will continue indefinitely (regardless of VMs added or moved at that level).

The main downside with virtualization backup tools is that many only back up virtual servers. The other drawback is that most of them don't offer any solution for getting backup data off-site, other than replication. In other words, sending backup data to tape is not an option, so if you don't want to do replication -- typically because you don't have a secondary location or because the bandwidth is too expensive -- then you don't have a way to get your backup data off-site in the event of a disaster.

So while virtualization backup tools have a number of unique features, they lack important functionality for those who still have traditional physical servers.

David Davis
is a virtualization evangelist at, the global leader in video training for IT pros. He holds several certifications including VCP, VCAP-DCA and CCIE #9369. He has been awarded the VMware vExpert award three years running. Davis has spoken at major conferences like VMworld and written hundreds of articles for websites and print publications, mostly around virtualization. He has authored 10+ video-training courses at TrainSignal including vSphere 5, vSphere Troubleshooting, and vCloud Director Essentials. His website is You can follow him on Twitter or connect with him on LinkedIn.

Dig Deeper on Virtualization and private cloud