ClickHouse Meets Postgres: QuintoAndar's Universal Data Bridge

Bridging the Data Gap

QuintoAndar's adoption of ClickHouse Managed Postgres to integrate with tools not natively supporting ClickHouse is a compelling case study. The core innovation lies in leveraging the pg_clickhouse extension to expose ClickHouse's analytical prowess through a familiar PostgreSQL interface. This effectively transforms ClickHouse into a 'plug-and-play' data source for any Postgres-compatible application, as highlighted by its seamless integration with Hightouch. The elimination of a complex Lambda architecture in favor of a unified ClickHouse-centric approach, coupled with the ability to retain existing tools like Hightouch, demonstrates a pragmatic and cost-effective solution for scaling data platforms. The article effectively showcases how this managed Postgres layer acts as a universal adapter, preserving investments in both analytical performance and existing third-party services.

However, a deeper dive into potential limitations and long-term implications would enhance the analysis. While the article states the performance impact is negligible for Hightouch's use case, the overhead of the double hop (Postgres to ClickHouse and back) for other, potentially more latency-sensitive applications, warrants further investigation. The security model, while mentioned as being managed through user roles, could be elaborated upon to understand how granular access control is enforced across this hybrid setup. Furthermore, the reliance on pg_clickhouse suggests a dependency on this specific extension; understanding its maturity, maintenance, and potential future compatibility with evolving ClickHouse and PostgreSQL versions would be beneficial for long-term strategic planning. Despite these points, the solution presented by QuintoAndar represents a significant step forward in democratizing access to powerful analytical databases like ClickHouse, making them more accessible to a broader ecosystem of tools and developers.

Key Points

• QuintoAndar consolidated its customer data platform (CDP) from a costly two-pipeline Lambda architecture to a single pipeline using ClickHouse Cloud.
• The previous architecture involved separate systems for low-latency reads (Postgres) and batch processing (Databricks), leading to complexity and cost.
• ClickHouse Managed Postgres, utilizing the pg_clickhouse extension, acts as a universal interface, allowing Postgres-compatible tools like Hightouch to query ClickHouse seamlessly.
• This solution enables ClickHouse to be integrated with tools that do not natively support it, effectively making ClickHouse 'plug-and-play' for a wider ecosystem.
• The architecture simplification reduced maintenance overhead, eliminated expensive duplicate event processing, and improved data freshness.
• Hightouch can leverage ClickHouse's analytical power while persisting its own state information in the managed Postgres layer.

---

📖 Source: A universal interface: How QuintoAndar made ClickHouse plug-and-play with managed Postgres