Logical Replication in Postgres
Logical replication in postgres: What is logical replication? When to use logical replication? Architecture of logical replication:
- Publications
- Subscriptions
- System tables/views
- Logical replication statistics
- Restrictions
- Current work
- Future work
The PostgreSQL logical replication follows the publish/subscribe model. The publisher node creates a publication, which is also called a change set or a replication set. The publication is simply a set of changes from one or more tables. The subscriber node creates a subscription, with the capability to subscribe to either one or more publications.
The logical replication in PostgreSQL follows the steps given below:
Step 1: A snapshot of the publishing database is copied to the subscriber. This step is also called the table synchronization phase. To reduce the amount of time spent in this phase, you can spawn multiple table synchronization workers. However, you can only have one synchronization worker for each table.
Step 2: After the copy is completed, subsequent changes made in the publisher node will be sent to the subscriber node. The changes will be sent as they happen in real-time. The changes will be applied in a commit order to ensure there is transactional consistency.
