Open data endpoints

Open data endpoints#

Every virtual database in Querona can be published as a modern API — REST, GraphQL and MCP — generated automatically from your existing objects, with no API code to write. These endpoints sit alongside the SQL Server (TDS) endpoint, so the same governed data is reachable by applications, web and mobile clients, partners, and AI agents.

Powered by Data API Builder#

The REST, GraphQL and MCP endpoints are served by Microsoft Data API Builder (DAB) — an open-source (MIT) engine that turns database objects into full-featured APIs from configuration alone, with no hand-written CRUD code. Querona embeds DAB version 2.0.9; for the general capabilities of the engine — the REST and GraphQL feature set and the MCP tools — see Microsoft’s Data API Builder documentation.

This guide covers what Querona adds on top: it generates the DAB configuration automatically from your virtual databases and the permissions you have defined — so the API contract follows your model and your access rules with no configuration files to author — and it manages the endpoint instances from the dashboard (see Managing Data API).

Two ways to run it#

Built in (default). DAB ships inside Querona. On start, Querona generates the configuration and runs DAB as a managed process — no extra infrastructure to stand up. This is enough for small or experimental deployments.

Standalone, with Querona as the data source. For independent scaling, or to fit an existing DAB deployment, you can run DAB separately (for example as a container) and point it at Querona. On its own, DAB connects to a fixed set of engines — SQL Server, Azure SQL, Azure Cosmos DB, PostgreSQL and MySQL. Because Querona presents a SQL Server-compatible endpoint, DAB sees it as SQL Server — which lifts that limitation: through Querona, DAB can publish data from any source Querona federates (Excel workbooks, CSV and Parquet files, REST and SaaS APIs, big data, and more) with no ingestion or copying. A separately hosted DAB is managed outside Querona.

What you get#

One definition, three protocols — REST, GraphQL and MCP from the same objects, at the same time.
No API code — DAB replaces most hand-written CRUD APIs.
Rich queries out of the box — filtering, sorting, pagination, column selection and relationship navigation, with an OpenAPI/Swagger description for the REST surface.
Secure by default — consumers reach data through the entity layer, never the database directly, and your Querona permissions gate every operation across REST, GraphQL and MCP alike.

Common use cases#

Back-ends for web and mobile apps, without building a bespoke API.
AI agents and copilots over MCP (see Querona for AI).
Headless or partner data sharing over a stable, governed contract.
Rapid prototyping against real, governed data.

AI agents over MCP#

Instead of letting a model write SQL against your systems, the MCP endpoint gives agents a small set of typed, governed operations: reading, filtering and aggregating records, calling the functions and procedures you exposed, and — where the granted role permits — creating, updating or deleting records. An agent can only use the operations you granted, on the entities you exposed, so there is no free-form query for it to get subtly wrong. The exact tools shipped with the embedded engine version are listed in Microsoft’s SQL MCP Server documentation. Why agents need live data — shown end to end on a banking example — is covered in Querona for AI.