But Ronald, is that even possible? Yes — and it’s easier than you’d think, now that we have SQL Server 2025 CES (see: [dbaronald.nl/just-say-yes-to-sql-server-2025-ces](https://dbaronald.nl/just-say-yes-to-sql-server-2025-ces/)).
With this app you get close: export a big table to Parquet, land it in Fabric, and do your analytics there. After an automatic upload to a Fabric workspace, we check it with Spark. When rows change or get deleted, use DuckDB to diff the old and new Parquet snapshots — added, removed, and changed rows, all in one query — so you know exactly what moved without re-exporting the whole table.
Is it real log shipping? No — there’s no transaction log, no restore, no recovery. Call it poor man’s log shipping. And here are the changes on the way to Fabric (though I’m pretty sure there’ll be an “ultimo 2”).
**Download:** [SQLtoParquet v0.1.0 for Windows](https://github.com/ronaldgithub/SQLtoParquet/releases/download/v0.1.0/SQLtoParquet-v0.1.0-win-x64.zip) — self-contained, no .NET install needed. Source and release notes: [github.com/ronaldgithub/SQLtoParquet](https://github.com/ronaldgithub/SQLtoParquet/releases/tag/v0.1.0).
