Would I be wrong in thinking that this "bi-directional logical replication" is only usable in a very limited number of use cases? As you mention, unique constraints pose a problem, and a lot of tables would have unique constraints to protect business keys?
The article references another article on this, the refers to the PostgreSQL documentation:
- conflicts [1]
- restrictions [2]
You need to be very aware of the limitations to decide whether this is usable in a specific context. I don't think this is really intended to be used for "bi-directional" logical replication.
This feels like something for the specific use case of heavy writes that are treating the DB like an append-only log, and then running analytics queries over the top.
Something like a distributed metrics collection system.
The article references another article on this, the refers to the PostgreSQL documentation:
- conflicts [1]
- restrictions [2]
You need to be very aware of the limitations to decide whether this is usable in a specific context. I don't think this is really intended to be used for "bi-directional" logical replication.
[1] https://www.postgresql.org/docs/current/logical-replication-...
[2] https://www.postgresql.org/docs/current/logical-replication-...