Data replication refers to a collection of technologies that offers alternatives to traditional backup and recovery techniques, and to traditional archiving. Generally speaking, replication technology falls within one of three categories: snapshots, volume copies and remote mirrors. Used individually, these technologies can be extremely helpful. Used collectively, they provide an extremely powerful answer to a number of vexing IT problems.
How does replication get used? Here are some scenarios.
First, create multiple volumes within a single array and put high-value data - a sales database, perhaps - on one of the volumes. Next, clone the database (the "prime" volume, which is the production system), creating an identical dataset on a second volume within the array. Volume copying software lets you do this with very limited impact to the production system.
Step two is to use snapshot software, which creates point-in-time copies of the database. Each snapshot captures all the changes to the primary volume made since the previous snapshot. Because the system is capturing only delta blocks, very little data actually moves between the two volumes, so impact on performance is kept to an absolute minimum.
What's the value of all this? The secondary system now has a data set that is identical to the primary volume, updated by the snapshots according to whatever level of granularity the system requires. When data corruption or any other problem occurs on the primary volume, your recovery software merely initiates a disk-to-disk recovery rather than trying to recover from tape. In just about every case, this will be faster than retrieving data from tape.
Having a secondary system exactly mirroring a production system also means that backups no longer have to be done off the primary, which will be of enormous value to any admin who has to provide 24-7 up time or is contending with a back-up window that has no flexibility in it to accommodate any problems in the back-up process. Backing up from the replicated volume means that even while backups are going on, production systems are unaffected and users never see any performance degradation.
That is not the only advantage to having an up-to-date mirror of production data, of course. Having replicated data means that IT now can test systems using real-world data configurations before they are put into production. Upgrades can be tested in environments that closely approximate production environments.
Data replication requires disk space of course, and the technology shouldn't be applied willy-nilly to all the data within the data center. Managers able to identify which datasets are business-critical, and who are building IT systems to economically manage information lifecycles will find this to be a cost-efficient alternative both in terms of IT demands and business requirements.