Maintaining a backup and recovery process when you manage remote offices is challenging. Typically, each office has one or more file servers to back up to tape, which is easy enough if you have an IT staff person on site but tricky if you have to beg a non-IT person to periodically change tapes for your smaller offices. Even worse, walking a non-IT person through a remote office restore process is like trying to drive the Los Angeles freeways blindfolded during rush hour.

Replication software, which is normally used for geographically distributed clustering, is a reliable technology. In the June 2002 Windows & .NET Magazine article "Geographically Distributed Clustering, " Ed Roth compared three replication solutions, from Computer Associates (CA), Legato Systems, and NSI Software. Roth found that all three solutions work well for replicating data across a WAN and provide failover capability. But you can also use replication software as an enhanced alternative to the typical remote tape backup solution.

Suppose you have 10 remote offices throughout the United States that are all connected to a central office in Cleveland through a VPN. At each remote office, you load a copy of replication software on the server. At the central site, you load a copy of replication software on a high-end Networked Attached Storage (NAS) server. Once configured, the software replicates changes on the remote servers to the central server in realtime. The volume of changes and the bandwidth of your VPN line will determine when your central server has a complete and current copy of each of the remote file servers. Because these copies are only of the remote data, you could back up the central server to tape at your leisure without worrying about a backup window and without the hassle of managing remote office tape libraries. And, at only $2500 per server, buying replication software is cheaper than having to buy a good tape device and tapes for each remote office. Best of all, if any of your remote servers go down, you can configure your central server to take over remote operations until you can recover the remote server. Remote office workers could get their data directly off the central server through the VPN.

Typically, a disk failure in a remote office requires a file server restore; end users lose changes they made after the last tape backup, which is typically the night before the failure. If the failure occurs late in the business day, an entire day's work can be lost. With replication software, you can reduce data recovery time to 15 minutes or less.

For even more thorough disaster prevention, you can create a complete copy of the central server at another hot site to replicate the central backup server to a remote location in, for example, New Jersey. If the entire central site goes down, you can reroute the remote office servers to a central backup disaster-recovery site in New Jersey. This scenario provides maximum data protection and availability to remote users' data.

To find out more about the solutions mentioned in Roth's article, visit:

CA at http://www3.ca.com

Legato at http://www.legato.com

NSI Software at http://www.nsisoftware.com