The growth of virtualised server environments is allowing more flexible and cost-effective server environments than ever before, not to mention that it is a major technological element which goes into what we know as ‘the cloud’.
Given this is a critical piece of infrastructure that many, if not the majority of businesses rely on, our engineers at Kroll Ontrack see common mistakes and oversights when performing virtual data recoveries on these systems.
Extreme server consolidation leading to a single point of failure.
We have had to deal with customers whose entire IT infrastructure was down due to a single system failure. You can normally tell that this has been the case when the email address they provide ends with @hotmail or @gmail!
Bad virtual machine lifecycle management
We often find that companies have failed to document the creation of virtual machines. Without having a procedure in place for creating and decommissioning virtual machines it is very easy to end up with a bad case of virtual server sprawl.
This has negative consequences for the virtual environment while it’s operational (waste of resources on unwanted VMs, risk of not backing up critical required VMs, risk of deleted the wrong VMs, etc), but it’s also problematic when there is a need for virtual data recovery:
- No one knows what the names of the critical virtual machines are. Normally we find several copies of a VM like ‘SQL-Server1’, ‘SQL-Server-Copy1’, ‘SQL-Server2’ etc, and no one knows which ones was the VM in production and which ones were tests or old versions of it.
- No one knows how the virtual machines were configured, how many Virtual Disks (VDs) they had, how the VDs were provisioned (thick or thin), what was the capacity of each VD etc.
- Virtual Disks may have ended up at a different storage device to where the virtual machine configuration is.
In the end it means having to recover a lot more data than it’s really required just to be able to find the right VMs.
Misunderstanding of snapshots
People misunderstand snapshots and their purpose, many times using them as a means to backup data. Misuse of snapshots normally leads to storage devices running out of space, and when that happens people start deleting snapshot files thinking they don’t need them anymore (thinking they are some form of backup). Snapshots files in virtual environments contain the most recent data as opposed to snapshots in SAN storage systems where they contain old versions of the data.
Some companies have not adapted their backup strategies to benefit from virtualisation and in the worst cases virtualisation plays against their outdated backup strategies. We have done many recoveries where customers could not restore data from backups because at a lower level the backups shared the same storage system as the production data and the cause of data loss affected both production and backup data (another example of over-consolidation).