It is intended as a long-term data relationship between databases.Typical uses are in: There are three methods of replication: Snapshot, Transactional, and Merge.These are provided to try to meet the wide range of business requirements for replication.
Replication is intended to be a way of distributing data automatically from a source database to one or more recipient databases.
As such, it can have obvious uses in a distributed system.
It has also been used to implement high-availability systems.
It is not useful for one-off synchronization, or for simply copying data.
It holds the definition of the 'Publication', which defines the 'articles' that are to be 'published'.
(The database with the original location of the data determines what is to be distributed).A 'Subscriber' receives the articles from a publisher. Any database can take on either role or even both roles at once.An Article is the smallest component of distribution, and can be a table, procedure or function.If it is a table, then a filter can be applied to it so that only certain rows or columns are replicated.This 'Magazine Publishing' analogy can be confusing because, in replication, a Subscriber can sometimes make updates, and a Publisher usually sends out incremental changes to the articles in a publication.A 'Publisher' maintains the original copy of the data.