SQL Server's database replication function lets database changes propagate across a LAN or WAN connection to one or more target systems. SQL Server's heterogeneous replication capability even allows replication to other database platforms such as Oracle, DB2, and Access. Here's a look at SQL Server's seven core database replication components.
An article is the basic unit of replication. The article represents what's published. An article can be a database object, such as a table or a stored procedure, or a selection of table columns and rows. SQL Server can't replicate an article unless it's part of a publication.
A publication consists of a group of one or more articles. A publication represents a data set that SQL Server will replicate. In other words, SQL Server replicates the publication, not individual articles. Publications let you group related data and objects for replication.
The publisher makes the publication available for replication. A given article can only be published by one SQL Server publisher. The publishing system contains the master replication database, which holds the data to be published. With some exceptions, only the publisher can modify the data that's being replicated. The publisher retrieves the information that will be replicated from its database and forwards the information to the distributor.
The distributor is the SQL Server system that's responsible for transferring the replicated data from the publisher to various subscribers. For small-scale replication, the distributor and the publisher are usually the same system. For large-scale replication, organizations implement the distributor as a separate system.
The subscriber is the system that receives the replicated publication. Whereas a SQL Server system has only one publisher, it can have multiple subscribers. At first, this design might seem limited, but a publisher of one publication can also be a subscriber of another. Likewise, a subscriber of one publication can also be a publisher of another publication.
2. Push Subscription
In a push subscription, the publisher sends the publication to the subscriber with- out waiting for the subscriber to request the publication. Transactional replications use push subscriptions because the replicated data must be available in near realtime.
1. Pull Subscription
In a pull subscription, the subscriber periodically requests the publication from the publisher. You use pull subscriptions when you have many subscribers and no need for immediate availability of the replicated data from the subscriber.