There is a lot of hype in the marketplace about virtual machine backups at a host level to backup infrastructures in its entirety in order to deploy it elsewhere as part of their backup and disaster recovery (DR) strategy. In reality, these types of backup and DR strategies should not be categorised together, says Gareth Tudor, CEO at Altonet.
DR is required in the event that an organisation suffers a complete loss of infrastructure. While it is paramount to plan for such an incident, in by far the majority of cases, companies only experience a partial data loss, such as deleted files, missing log files on only one server as opposed to their entire data centre.

Because virtualisation is still relatively new in the SME marketplace, backing up entire servers is a knee-jerk reaction because it is touted as the easiest way to recover from a total data disaster.

Many SMEs have not thought their DR and backup strategies through. Virtual machine backups as a primary function is inefficient and operating system (OS) level backups give them similar, if not the same functionality to host level backups.

However, it requires a bit more effort and planning, but gives companies the same granularity, which is the ability to restore a single file, rather than restoring the entire drive as well as efficiency of data.

When doing virtual machine backups, companies essentially backup their virtual disks in their entirety, and are shipping entire drives across the network to the DR site.

There are distinct advantages to that strategy because with virtualisation, the virtual hardware is extracted from the physical hardware, so there are a lot less issues when moving servers between physical hardware, and if the servers are platform agnostic the process is relatively painless as long as the new server runs on the same platform.

On the other hand, if they backup on OS level, certain functionalities, such as incremental backups come in to play, which is a much more efficient way of doing things.

Companies can do “bare-metal” restores which restore not only the data but also the OS, applications and data. Therefore, in a total server loss scenario, OS backups can be re-provisioned and the machine can be restored to its original state or when the last backup was taken.

Another advantage of OS backups is that companies are able to include or exclude certain files when they backup, as opposed to a virtual machine backup. Files such as emails that need to be stored for compliance and governance purposes can be backed up regularly, as opposed to transient files and data.

This also makes maintenance of servers much more efficient. From a timing perspective, this also makes backups much quicker and efficient.

Tudor’s suggestion to SMEs for an efficient backup and DR strategy is the following: if they have a bottomless budget, they can use virtualised backups and OS backups, as they would ideally want to restore single files if need be, but with virtualisation users have a DR strategy as well.

However, in the real world where tight budgets are a reality for smaller businesses, OS level backups would suffice even though the process takes a bit longer.

That said, backup and DR is not the same thing. Backup is leveraged for DR, but just because you have a backup, does not mean that you can successfully recover from a disaster. That is where virtualisation comes in, which makes recovery a lot quicker.

This brings me to another point: recovery time objectives and budgets. How long after a data loss do users need to be up and running? If a week is sufficient after a complete data loss, OS level backups will be sufficient and cost-effective. If you can only afford a day, virtualised backups are the way to go, but at a price.

In closing, for most SMEs Tudor believes that from a cost and recovery time perspective, OS level backups provide a good value proposition, because it gives you the ability to recover from a disaster, but requires a bit more planning.