Overview

In production settings, we recommend deploying ParadeDB as a replica of your primary Postgres, connected via logical replication. This allows ParadeDB to integrate with managed Postgres providers like AWS RDS, Google CloudSQL, and Azure Postgres. It also isolates search and analytical workloads from your primary Postgres, which can prevent downtime caused by long-running queries.

​

Logical Replication

Logical replication is a tool, native to Postgres, which enables keeping two or more Postgres databases synchronized in near real-time in a fully transaction-safe way, without needing an ETL or CDC pipeline (e.g. Debezium). It is possible to ingest a subset of tables or rows/columns via logical replication, and it is possible to ingest from many primary Postgres databases.

More information on how Postgres logical replication works can be found in the Postgres documentation.

​

Architecture

If your primary Postgres is hosted by a managed Postgres service, we recommend deploying ParadeDB on Kubernetes. To configure replication, please refer to documentation for your managed Postgres service.

If your primary Postgres is self-hosted, we recommend installing the ParadeDB extension(s) directly in a replica within your cluster.

​

Development

In development, we recommend deploying ParadeDB with Docker.