ClickHouse as a streaming HTTP API

ClickHouse 26.8 introduces named HTTP handlers that allow engineers to define URL endpoints directly in the database. Each handler maps to a SQL query with typed parameters, enabling HTTP requests to trigger database operations without an intermediate API layer. This change targets architectures where a thin proxy currently translates HTTP parameters into ClickHouse queries and returns results. The feature supports pagination, framing formats like JSONEachRow, and access control via ClickHouse's user permissions system.

The operational cost of this approach is reduced complexity for query-only APIs. Engineers no longer need to maintain a separate service for simple data exposure, and the database handles streaming responses directly. However, this shifts responsibility for query safety and performance to the handler definitions. Poorly designed handlers could expose inefficient queries or sensitive data if access controls are misconfigured. The feature does not replace application layers that require business logic, orchestration, or custom validation beyond what ClickHouse's SQL dialect supports.

Where this approach stops working is in scenarios requiring dynamic query construction or complex request processing. Named handlers are static definitions, so each new query variant requires a new handler. The feature also lacks built-in rate limiting or request validation beyond basic parameter typing. Engineers must still implement these protections elsewhere if needed. For use cases that fit within these constraints, the feature offers a simpler architecture for exposing ClickHouse data over HTTP.