Understanding BDR replication concepts

Back in the old days, way before 9.0 was introduced, people had to use Slony to replicate data. The core problem with solutions such as Slony is that there is a need for a changelog trigger, which actually writes data twice. Trigger-based solutions are hard to administer, are not able to handle DDLs, and are in general a bit tricky to operate.

BDR has been invented to put an end to trigger-based solutions and turn PostgreSQL into a more robust, more scalable, and easier way to administer solutions. Trigger-based replication is really a thing of the past, and should not be seen in a modern infrastructure anymore. You can safely bet on BDR—it is a long-term, safe solution.

Understanding eventual consistency

In the early parts of this book, the CAP theory was discussed. It is a vital ingredient, which should always be kept in mind when a new database technology is evaluated. In the case of BDR, the system is eventually consistent. What does this...