I’ve heard several of the arguments in favor of a server that supports datareplication. What are the potential problems?
For our readers, the advantages of data replication [creating copies of data, orfragments of data, at one or more sites] include the improvedperformance that comes from working with local data, and the increasedavailability of data, which is especially important for a global operating environment (data can continue to be retrieved even if the primary site becomesunavailable provided that a copy continues to be available).
The major disadvantage is that every copy of data that has been replicated also must be updated. (If one or more copies of the data do not agree, data integrity has beencompromised.) Under ideal circumstances, one would want to have all primary and replicated data updated simultaneously, thus avoiding loss of integrity. Unfortunately,this would potentially defeat the major benefits of replication ? if a site becomesunavailable, updates cannot be replicated there, which would in turn cause the transaction to fail (the same result if the entire system were unavailable). To avoidsuch a negative scenario, most commercial systems that support data replication propagate updates on a delayed basis. Thus, integrity that is taken for granted on an unreplicated system is sacrificed with delayed replication. Most businesses that need data replication are able to work within this limitation, but it obviously is a limitation that should be known beforehand.