Replication is the process of creating and maintaining a separate copy of database in another server or the same server and keeping both synchronized. Replication is useful in a number of scenario:
1. To integrate multiple database to build a central warehouse: Businesses operating on a number of distant locations usually collect data in separate database. However the senior management needs to generate reports showing analysis of data collected from all these sites. Hence the need for central warehouse is increasing and replication proves to be a powerful way to create and maintain the warehouse. It can also be used to create and maintain a separate data mart for OLAP operations.
2. To ensure higher availability and improved performance of web applications: Web applications with the increasing number of users faces huge problem in avoiding the extra load on the database. Replication along with other techniques can offer the right solution by distributing users to a number of replicated database in different servers and thus improving the performance and ensure high availability.
In Inmon's architecture data from OLTP databse is stored in a warehouse before it is transferred to the data marts. In Inmon's view the data warehouse is a real database.
According to Kimball's architecture data is transferred directly from OLTP database to the data mart. In Kimball's model the data warehouse is actually a collection of integrated data marts.
