Planning for distributed disaster recovery

Many businesses have multiple locations, which leads to data being stored on multiple file servers in those locations. There are many ways to keep that data safe, but branch office--or other forms of distributed office--Disaster Recovery (DR) can lead to some interesting problems.

The problem with tape backup
First off, while tape is a form of DR that can be used for branch operations, keep in mind that there will be a steep learning curve if you're depending on nontechnical branch staff to use the tape backup. Most branch offices have no technical staff to deal with the intricacies of managing and tracking tapes. The task is usually delegated to those who may be exceptional at other business tasks, but have no idea how tape drives work, such as administrators or branch managers. Over the last several years, I've heard just about every horror story: from the admin assistant who put the cleaning tape into the drive every other night, to the branch manager who used the same tape for four years. Needless to say, none of these tapes worked properly when it came time to restore the data that had been accidentally lost, so the DR plan was not very successful.

Automated systems: RAID and VSS
Automated systems can help to remove nontechnical staff from the equation, but keep in mind that these systems won't allow you to move the data off-site in most cases. For example, you can use a RAID array to keep the data on a storage system that can withstand the loss of one or more drives. You can also use Volume Shadow Copy Service (VSS) from Microsoft or third-party tools commonly available on the market. Both of these types of systems will allow you to recover from a local disaster that doesn't involve data corruption. The benefit is that both of these automated solutions can be set up and monitored remotely, thereby moving the administration of backups to an office where you have technical expertise.

Replication
The next level of dealing with branch office backups would be to actually move the data back to a central office in order to back it up to tape where you have tech staff ready to keep an eye on the backup systems. This can be accomplished either with hardware-based systems that are all-in-one packages for data storage and replication, or by using host-based tools designed to move data from server to server instead of disk to disk. Both types of systems can allow you to move only changes to data, but depending on the size and scope of the projects, you may find that hardware-based systems end up being less cost-effective in smaller implementations.

No matter what kind of replication you use, nearly all these systems will allow you to take a snapshot and back up your data at centralized sites, eliminating the need for nontechnical staff to manage your backup systems. Keep in mind you may need to move large amounts of data back over lower-bandwidth links in the event of a total restore. You can get around this by keeping a standby server at the central site, moving the data to it over high-bandwidth LAN links, and then shipping the server to the failed site. That site sends the failed server back to you, where it's repaired to become the standby server for the next disaster.

No matter if you train local staff, or get the data back to technical staff for backup, you must protect branch offices against data loss and disasters. Knowing what options are available can make it easier to show management that just because the data isn't under the same roof, doesn't mean it isn't important.

Mike Talon is an IT consultant and freelance journalist with several years of experience in disaster recovery and high availability technologies.