Chapter 4. Concepts
dbreplicator is a database replication/synchronization tool that synchronizes data between multiple heterogeneous database backends including servers, desktop databases and mobile databases. It uses a publish/subscribe model much like a standard server/client.
It has been engineered to perform bi-directional, multi-master replication between databases such as Oracle, SQL Server, DB2, Daffodil DB, PostgreSQL, MySQLand Apache Derby, enabling high-performance distributed database connectivity, with databases residing on Windows, Linux, UNIX or Solaris. These databases can be on servers, desktops or mobile devices. Filters allow administrators to synchronize all or only a subset of database records.
This enables users to work against a local copy of a database and then synchronize changes with a master or publisher server across a local or wide-area network.
4.1. Replication Terms
- Publisher
The passive replication member, generally the central server. A publisher is passive, listening for subscriber connections and then following their instructions.
- Subscriber
The active replication participant, generally at the satellite location. This connects to a publisher and initializes a replication session.
- Snapshot Replication
Takes a "snapshot" of the database from the publisher and moves it to the subscriber. This one-time process is a pre-requisite of Merge, Push and Pull replication.
- Pull Replication
compares Publisher's and Subscriber's data and the changes made to the data by the Publisher are updated in the Subscriber's database. But the changes made by the Subscriber are not updated in the Publisher's database. So in Pull replication, records (that are changed due to insert, update and delete queries in Publisher's database) are pulled from Publisher to Subscriber by the Subscriber. Before attempting Pull replication (for the first time), it is important to take the snapshot (a one-time process) of data from the Publisher. Once the snapshot is taken, you can perform Pull replication any number of times.
- Push Replication
compares Publisher's and Subscriber's data and then the changes made to the data by the Subscriber are sent to the Publisher's database, but the changes made by the Publisher are not pulled back to the Subscriber's database. So in Push replication, records that are changed due to insert, update and delete queries at Subscriber's database are pushed to the Publisher by the Subscriber. Before attempting Push replication (for the first time), it is important to take a snapshot (a one-time process) of data from the Publisher. Once the snapshot is taken, you can perform Push replication any number of times.
- Merge Replication
Merge replication takes changes to data from the Publisher and Subscriber and updates the changes in both databases, as though a "pull" followed by a "push" were executed. Before attempting merge replication for the first time, it is important to take the snapshot (a one-time process) of data from the Publisher. Once the snapshot is taken, you can perform merge replication any number of times. Merge Replication is also known as Change-based Replication or Synchronization.
 | Although a publisher can have many subscribers, a subscriber can only be subscribed to one publication. |
