Deep virtual protection is a phrase Unitrends coined to describe the advantages of being able to protect at both the hypervisor level (as Veeam and others do) while also being able to protect deeper into the VM (virtual machine) at the application- and operating system-level.   In essence, we believe that the ability to not only recover at a granular level but also backup, replicate, and perform other operations at a granular level is an important feature for any continuity and data protection solution.  This enables a virtualization-oriented administrator to more fully virtualize their environment rather than virtualization being stalled due to data protection challenges.

At the heart of the concept of deep virtual protection is the fact that all VMs are not created equal.  The average VM is well-protected by hypervisor-level-only virtual protection.  However, there are VMs with unique characteristics (such as larger VMs, VMs with high change rates, VMs requiring longer retention, etc.) for which a one-size-fits-all hypervisor-level-only data protection scheme just isn’t flexible or agile enough.

The most common user benefits for deep virtual protection include:

Applications and operating systems on VMs with high change rates.  We have a lot of Veeam customers that come to us asking for help with a VM with an application such as Exchange that has a high change rate.  Every time a Veeam snapshot is taken, the application freezes for a few seconds to a minute or more.  When this occurs, we use deep virtual protection to protect the application and the operating system at a native level – which means that we can achieve the same or better RPO (Recovery Point Objective) with a much less impact since we’re operating at the application transaction level.  Another common example are VMs with high change rates in some set of files or folders; in this case we use deep virtual protection with granular operating system and file system capabilities to either protect that data on a different schedule or to exclude it entirely depending upon the needs of the customer.

Applications and operating systems on VMs that require lower local RPOs (Recovery Point Objectives.)  This use case is the cousin of the change rate issue discussed above.  Veeam extensively advertises 15 minute RPOs – just as Unitrends advertises RPOs as little as one minute.  But that’s just the best-case period between scheduling backups (or replications, or other protection capabilities.)  What it doesn’t account for is how much data must be backed up and the speed of the underlying production system as well as the continuity and backup solution.  Unitrends offers deep virtual protection with granular backup that allows different applications, file systems, and files to have different accompanying backup strategies – as well as more efficient native-level backup.

Applications and operating systems on VMs that require better deduplication.  Deep virtual content-aware deduplication, or deduplication that understands not only the backup format but also the data being backed up, typically has a higher associated deduplication rate. See the blog post Unitrends and Veeam: Content-Aware Deduplication for a full description of this.

Application and operating systems on VMs that require lower replication RPOs and/or more granular WAN bandwidth management.  There should be multiple places from which you can recover in a continuity and backup solution – these include the vendor’s cloud (Unitrends Cloud; Veeam Cloud does not exist), an MSP (Managed Service Provider) cloud (both Unitrends and Veeam support MSPs), a second-site solution (which Unitrends and Veeam support), local or hyperscale cloud  archiving (Unitrends supports rotational disk, tape, NAS and fixed NAS and SAN as well as Amazon AWS, Google Cloud Storage, and other third-party clouds; Veeam supports tape), and local recovery from the backup pool (both Unitrends and Veeam support this of course.)  Deep virtual protection and granular backup allows the user to specify different backup strategies within each VM which then allows different recovery strategies at the intra-VM level.  This enables different RPOs to be set for granular backups as well as allowing resources within a VM to be protected via different archiving and replication mechanisms.  Or to put more simply, you can take a VM and most of it to tape but replicate Microsoft SQL to the Unitrends Cloud if you have relatively little WAN bandwidth to use for continuity and backup.

 

So with all of these benefits, when do we advise against using deep virtual protection?  For simpler virtual machines that have tight backup window requirements, hypervisor-level backup is faster because it doesn’t need to understand the underlying data.

So what do you think?  As always, we’d love to hear from you.

 

Guide