I’ve been mulling a post about this in my head for a week now and I thought I would get some thoughts out there on what it means to verify your backups. Veeam’s marketing department has done a great job in getting the word out about SureBackup but what is it really? Some have suggested it’s nothing but a patch for our synthetic backup approach. Some have even gone so far as to indirectly say through anonymous corporate blog posts that we have a poorly designed product with Backup & Replication version 4.0. I guess not everyone “gets it” so that’s why I’m writing this post.
What we’re talking about with verification has nothing to do with how we store our backups but everything to do with the virtual machines inside of our backups. Our synthetic backup approach is a proven technology that we did not invent, simply refined. Our source based de-duplication of data as well as removal of “empty” blocks has been proven for over 2 years in production environments. So if you think all of this buzz is just about verifying the integrity of the “backup file”, think again.
Veeam R&D did not initially set out to provide verification capabilities. We were actually doing research into how to provide functionality that our customers were asking for, namely application-item level recovery. Since we are doing image-level backups, how could we allow our customers to retrieve individual application items such as email, SQL, etc.? Some vendors recommend backing up data twice, once with an image-level product and again with a different agent based product (with specific agents for each application). This approach means 2 backup vendors (not integrated), resource contention on hosts and almost double your backup space…we knew there had to be a better way.
Building on our advanced file-level recovery capability we knew that if we could just start a recovered VM we could then start the application and pull out individual application items. In the past this was done by restoring the entire VM in an isolated network, a process that can take a considerable amount of time and effort for just 1 email. What if we could just run the VM from the backup file without having to extract it? R&D quickly got to work to see if this was even possible…
The Aha Moment
Sometimes in software development you start out to solve one problem and in the process realize that you can solve other problems, problems that no one ever thought could be solved. When was the last time you started a server running from a tape backup? When was the last time you verified the ability to successfully recover (and start) all of your servers that have been backed up (monthly? quarterly? yearly? never?)?
Veeam R&D quickly realized that they were on to something big here, what some call the “aha moment”…what if we could startup the VM’s directly from the backup file in an isolated environment just after the backup occurs? Could we actually give our customers peace-of-mind that their image level backups (which are simply crash consistent) would perform as expected if they needed to do a full recovery? The answer to those questions is “Yes we can”!
So that is how this whole verification idea came about, we wanted to provided item-level recovery and we ended up doing that as well as something thought impossible by many.
- Comprehensive – maintain dependencies across VM’s (email, directory, DNS) to provide the verification
- Flexible – allows custom scripts in verification jobs based on specific OS and application
- Fast – starts the VM’s, checks them and shuts them down…in minutes
- Non-destructive – No changes are made to what’s stored in the backup, the verification environment is fenced off from production
- Automatic – Once the jobs are created, they run on schedule
- Peace-of-Mind – once verified, you know you can recover the VM’s and they will work.
So when you think about verification, take the points above into consideration. Verifying the integrity of a backup file is easy, verifying the fact that the VM will work when it’s recovered? That’s not so easy…but it will be soon.