Idea for supplementary backup verification method

[christiankelly]

SureBackup is a great technology but can be complex for smaller environments with the networking requirements, boot sequences and scripts. I’m wondering if others feel an alternate “boot screen snapshot” verification would be useful?

Basically it would be a simple boot with delay, (set by user) which after the allotted time would take a snapshot of the boot screen, and add it to the email or logging. This would prove that the backed up VM could boot, and although not as in-depth as SureBackup it would give enough peace of mind for many clients.

Datto has something like this so it’s not a unique idea, but I’m wondering if this has ever been considered by Veeam and if others would find such a feature useful. I know with a backup roll out that we’re considering which will cover 100 clients having a simple screenshot verification as a test method of successful backups would be welcome.

[Gostev]

Depends on what is the goal here is. If it is to give your clients false sense of comfort about their backups, then this is a good idea. Otherwise, it is quite useless because the OS will boot fine even if 9 out of 10 VM disks are missing from the backup completely nor I remember a single VMware CBT bug that actually prevented OS from being booted.

[christiankelly]

I agree that SureBackup is a preferable method, but there are many servers which just have one disk so booting would be an acceptable test in a lot of cases.

In many environments SureBackup isn’t being used at all due to its complexity, so while I agree it’s a better method, I would like to see something simpler added which would help uncover many issues which may not be noticed at all.

[GarethUK]

Erm.. This awkward.

Ok. I actually have this working. I wrote a Powershell script to do this based on Surebackup runs. I know it works pretty well as I needed the functionality for a large not very uniform environment and used it to determine that I had a problem.

There are various issues with doing it which are not obvious. I went down the route of OCRing each screenshot to text (bit tricky to get right) then working out what the text says and determining what is most likely to be on screen. Most of the time it is simple but it can be impacted by the build you use - especially on Windows 2003. However, this may have become easier with 2008+ but that has other issues - I've not got all the issues resolved but I know how to.

There are also various optimisation which mean you can shortcut a load of the work making the whole thing faster. Given a bit more time and change in architecture I'm sure you can get something that works in real time as the surebackup job is running.

Regards,


Gareth

[vBalmer]

Wouldn't a basic VSB cover this with the ping test? If the OS boots and the screenshot shows it looks good, a ping test should also work and verify networking in addition to it booting. This doesn't require any special scripting or extra work, just a "next next next" all defaults config.

[GarethUK]

Depends..

As I said, we don't have a completely uniform infrastructure. Our network is a full /16 plus probably the same again in private address space divided up into /23 or /24. Last time I checked putting that lot into a virtual lab wasn't a goer. I'd like to be proven wrong.

We do use IP based checks via surebackup for important systems within smaller application groups. But when it comes to verifying a large number of VM's we have it starts to become far from trivial. THere are some inflexibilities with surebackup which make it quite painful in these circumstances.

I produce a report which includes various bit of information and will eventually include a screen shots per machine. This does reference the VMWARE tool status but this appears unreliable in about 25% of our estate.

Regards,


Gareth