Cloud-Native Data Without Starting Over

TL;DR

• 1

  The easiest first step is swapping the SQL driver: Nixon calls changing from deprecated System.Data.SqlClient to Microsoft.Data.SqlClient a near drop-in update that brings current investments in performance, security, and new features.

• 2

  SQL Server 2025 changes search behavior, not just version numbers: the new vector data type and REST-based model integration let a search for "wireless keyboard" return semantically related items, with distances like 0.4 for close matches and 0.75 for looser ones such as an ergonomic chair.

• 3

  Data API Builder can replace a lot of hand-written CRUD code: instead of maintaining REST endpoints, unit tests, and pipeline logic for common database operations, DAB uses a JSON config to expose tables, views, and stored procedures over REST, GraphQL, and MCP.

• 4

  SQL database projects become the schema system when Entity Framework goes away: Nixon shows how stored procedures, tables, and views can be compiled, source controlled, and incrementally published, including CI/CD deployment scripts, without relying on EF migrations.

• 5

  A single API surface can sit in front of mixed back ends: DAB can merge multiple config files so one app endpoint can route across SQL on-prem, Azure SQL, Postgres, Cosmos DB, or MySQL during cloud migration.

• 6

  SQL MCP server is just DAB with MCP turned on: the same secured database surface used by the app can also be exposed to GitHub Copilot or other agents through JSON-RPC, with separate permissions if needed.