Data replication for disaster recovery
Data replication offers a fast, inexpensive way to get back up after a major data loss on the primary system. However there are a number of options for data replication and they differ significantly in their characteristics for data recovery.
Data can be either mirrored or shadowed for replication. Mirroring involves replicating every transaction as it is made. Shadowing logs changes to the data and applies the log to the replicated data at intervals. Data replication for disaster recovery usually involves sending data to an alternate processing site. A synchronous system, which mirrors every transaction as it is committed, requires a data link capable of handling the bandwidth. If the link to the remote site is too constricted, it will degrade performance of the production system. Synchronous replication usually requires a fiber-optic link between the two sites.
An asynchronous system, such as one that is log-based, is less dependent on the speed of the data link, but data on the replicating system may lag behind the data on the production system. The result is a window during which all data will not be current. Or, worst-case, data could be lost or if there's a problem when the system is synchronizing. Typically such windows will be between five and 60 minutes.
Gartner Group has a research note titled "Disaster Recovery: Weighing Data Replication Alternatives" (Technology T-13-6012) available at its web site, which discusses data replication and available products.
Rick Cook has been writing about mass storage since the days when the term meant an 80K floppy disk. The computers he learned on used ferrite cores and magnetic drums. For the last twenty years he has been a freelance writer specializing in storage and other computer issues.