A survey of 500 enterprises worldwide by VMware has found that there are major issue around back-up and recovery in virtual environments, which may be slowing adoption of virtualisation technologies in certain areas.
Carried out at the end of last year, the VMware Data Protection Report 2010 found that 61% of enterprises who are currently using physical-based tools for backup and recovery will now change their approach specifically because of virtualisation.
Despite the inherent advantages of virtualisation technologies, such as virtual machines (VM) that can be built and deployed in minutes, the report found that in the areas of back-up and recovery, there were significant limitations.
For example, the survey found that performing a full recovery of a backed-up VM still takes nearly five hours-a relatively small improvement on the six hours required to recover a physical server.
“The reason for it [the recovery time] is that when you have a virtual machine, it is different to a physical machine. Each VM becomes a file on the server’s hard drive,” said Ian Wells, director, Northern Europe, Veeam. “In order to back-up that virtual machine, if you just back-up the file, you only have the file. If you need to get back anything within it, you have to restore the whole file. If you then want anything out of it, you have to start it all up and a do a load of work to make that happen to pull the information out of it. We would call that a two stage recovery.”
Veeam is a specialist in back-up and recovery for the VMware environment.
“The second way of doing it,” explained Wells, “is to do guest level back-up, which is putting a traditional back-up agent inside each VM and back-up all the files as if there were on a physical server. That gets you the data that’s inside the VM but does not give you the ability to restore the whole VM.”
When asked if this was a moot point due to the features of virtualisation platforms such as live migration and seamless failover, Wells said that this was true to a certain extent, but with caveats.
“Yes and no, recovery is needed less, as systems such as failover can cope with more, but it is still necessary. But this kind of failover can be expensive and complex to do and is usually only done by larger organisations for the top tier of applications.”
These limitations have led 44% of IT directors, says the report, to avoid using virtualisation for mission critical workloads due to concerns about back-up and recovery.
Wells explained that in addition to the problems mentioned with two stage back-up and recovery, long back-up windows were a concern.
“They’ll probably be having issues with achieving their back-up windows as those backups tend to run very long in the virtual environment and when it comes to restoring, there are still a lot of failures. So, if you can’t protect your core data, you absolutely won’t put your production tier applications into that virtual environment.”
According to the survey, nearly two-thirds (63%) of organisations experience problems every month, when attempting to recover a virtual server from a back-up.
Wells said that these and other issues must be tackled before virtualisation will gain wide acceptance for mission critical applications in enterprise.
He said that Veeam had been working within the VMware environment to solve these problems, such as developing a product based around a virtual NFS server which allows our backup images to be mounted on a VMware host without doing anything to them.
Ltd
