virtualization.info

After teasing the customers for weeks, Veeam introduces today SureBackup, a set of technologies that greatly improves its backup/restore product for VMware vSphere.

First of all, SureBackup is not the name of a single new capability. It’s an umbrella for existing and upcoming features of Veeam Backup & Replication. And while this may sound like a typical marketing operation to turn old things into something new, the reality is that SureBackup includes at least a brand new, remarkable technology.

Dubbed Recovery Verification, this is the Veeam attempt to solve one of the most challenging problems when dealing with virtual machines live backups: testing that each guest operating system and its application will work properly after recovery.

In live (or hot) backups virtual machines are copied while the OS is running but when the backups are restored the OS starts from a powered off situation. This may lead to inconsistencies.
The Volume Shadow Service (VSS) technology that Microsoft offers since Windows Server 2003 and XP helps to reduce the risk when protecting Windows virtual machines, but to work effectively it must be supported by the applications that are running inside the OS.
Without VSS-aware applications, live backups don’t guarantee 100% safe restores. And there’s no way to have such guarantee without testing every single backup with a restore operation, which is way too expensive and time-consuming for a large number of corporations.

Recovery Verification, to be included in Veeam Backup & Replication 5.0 in Q3 2010, solves this problem by restoring protected VMs in an isolated portion of the virtual infrastructure.
The VMs are powered on from their backup archive (no VM is duplicated and moved across the data center), and put in a temporary virtual network that has no connection with other vSwitches or physical LAN.
Of course, every changes happening inside the VMs during this restore phase is recorded in a delta disk that is discarded as soon as the verification is completed. The archived image remains in read-only status to avoid any unexpected modification of the system state.

During the verification, the VMs are placed inside the existing virtual infrastructure if there’s no spare hardware, “stealing” some resources of the production environment. To reduce the impact, this operation can be scheduled at any time: after the VMs backup, during the night, etc.
Veeam says that each verification only requires a few minutes but this precious capability may turn into a completely new dimension to consider during capacity planning.
Alternatively, Recovery Verification can be used to test the disaster recovery hardware that is available inside the company.

The most interesting part is what happens to the VM after its backup image has been restored in the virtual infrastructure. Veeam goes beyond just checking that the guest OS can properly boot. The product runs specific scripts to verify that applications inside the VM are back in an acceptable operational status.
Veeam will offer some predefined scripts for the most common applications, but the key part is that customers will be able to use their own custom scripts to verify any sort of application activity inside the guest OS.
Clearly, not everything will be verifiable as the network connectivity is limited and other application tiers will probably be unreachable.

Once completed its tests, Recovery Verification will shut down the VM archive image, discard the delta disk and send a report with the results returned by the built-in checks and custom scripts.

This goes well beyond the verification of integrity for a backup image that disaster recovery vendors usually do and it’s definitively an interesting way to leverage virtualization. Veeam will support it with any guest OS (not just Windows) and any application. It will even work with Backup & Replication 4.x archives.
But the company has some news about restore too.

Less innovative, but still a very welcome addition, SureBackup also comprises the capability to restore application-level items, like emails from an Exchange Server, users from an Active Directory LDAP server or records from a SQL Server.
Proper roles and permissions for each object will be recovered too.

Veeam supports this capability in every guest operating system supported by VMware, without requiring the installation of agents, and by using the existing backup process used so far.

The same feature was announced in July 2009 by Vizioncore, as part of future releases of vRanger Pro.
And at that time virtualization.info speculated that Vizioncore’s parent company Quest may leverage the technology included in existing solutions it already sells for the physical world.

It will be interesting to see how the two companies will compete on this feature because part of the Quest expertise in application-items restore comes from the acquisition of Aelita Software, in 2004.
After the acquisition, several Aelita Software employees left Quest to found Veeam. The list includes Andrei Baronov: CTO at Aelita first, R&D Director at Quest after the deal, and finally CTO at Veeam.