Data Recovery Digest

Virtual machines are an important part of running a modern business. According to Gartner, the firms they talked to reported 75 percent or higher virtualisation. Hypervisors have improved over the years, which means it has become easier than ever to set up physical servers and maintain them. This leads to better server utilisation and IT departments that can be responsive and flexible.

There are loads of hypervisors to choose from: Oracle, Hyper-V and VMware just to name a few. Whichever you use, it’s important to be aware of the negatives. To take a physical server and turn it into many physical machines necessitates the addition of a software layer. It makes the admin experience simpler, but it actually makes the entire architecture more complicated. This is because the hardware is hidden; admins can begin to struggle to know what physical system a virtual machine is using. This is troubling when it comes to data loss and recovery.

There are various things that cause data loss on a virtual machine. The first of these is metadata corruption. Metadata is vital in a virtual environment because the amount of layers that are needed to run the environment. Data can stop becoming available if the metadata is damaged because it cannot be accessed.

Second, hardware issues with the RAID can cause data loss. A good system will replicate the data across different physical drives, but tied into a single logical collection. RAID is great because it improves redundancy and data reliability. In practice in fragments the data across multiple disks, pulling them when requested by the user, but data loss isn’t unusual. This is because a modern RAID system will use deduplication and compression, meaning that data cannot be rebuilt by the RAID if it becomes corrupted.

IT administrators will know the pain of reformatting disks in virtualisation and keeping software updated. Badly implemented new software or patches can cause file or even system damage. Formatting problems also arise when thin provisioning is used, whereby the amount of space required by the system is allocated on the fly, which results in a fragmented environment.

Finally, and this should never be underestimated, user error is a big cause of failure. As virtualised environments become easier to deploy, organisations are skimping on essential training. Upkeep of a virtual machine is more important than being able to spin one up. If an element of the chain is destroyed then the entire architecture is going to fail.

An enterprise should not neglect the importance of backing up virtual machines. They are far more complicated to recover from than standard server deployments. While data recovery is possible, through specialised recovery software or firms, it shouldn’t be necessary.

If you do need to use data recovery, first try recovering at the physical storage level. The structure of the virtual machine might not be recoverable, but the data within should be. If that fails, try from the logical volumes of RAID. The RAID can look at the data spread across the virtual disks, if it’s still intact. Failing that, try the host file system level or the virtual file system.