Data replication is an increasingly common way of solving backup problems, including shrinking backup windows and the need for near-immediate failover and restores. However without some careful forethought, replication can actually cause more problems than it solves, according to Robert Brower of Adexis Consulting, a Columbus, OH, storage consultant.
"As we go through the planning process customers say 'replication', we say 'why'?" Brower says. Often the answers reveal that there are cheaper alternatives to data replication or that the application hasn't been sufficiently thought out.
Not all data replication is the same, Brower points out. In fact replication comes in several different flavors for different purposes. Even though several of those purposes can be described as 'backup' they're actually quite different in their requirements.
"Do you want to create a hot site [for disaster recovery], do you want to allow for localized backup or are you looking for some way to allow a batch file like process?" Brower asks. All of these are common uses for replication, but all of them have different implications for the replication hardware and software.
For example if the purpose of replication is to provide a hotsite that can take over immediately in the event of a disaster, the system needs to replicate more than files. It needs to replicate the system state as well. "If I didn't replicate the system state and all I replicated was the data volumes, I can't really do an automatic operation at the other end," Brower says. Similarly, if the purpose is automatic failover, the system also needs to be able to boot from the replicated system. A replication system which is designed to provide very fast restore of corrupted or accidentally deleted files, on the other hand, doesn't need to worry about system state.
Synchronous and asynchronous replication also have implications that need to be considered. Both have advantages and disadvantages. Synchronous replication means that the copies of the data are the same -- right down to the problems. "If I'm using synchronous replication and I get a virus, I've just replicated the virus," Brower points out. "If I'm doing asynchronous replication and the connection breaks, my primary system keeps running and I'm keeping a journal of changes. But that means I need to allow for other disk resources to hold the journal."
Replication is a valuable technique that has many applications in storage. But it's important to understand the choices and trade-offs involved, Brower says.
For more information:
About the author: 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.