Tutorial: SQL analytics endpoint editor¶
CSA Loom
sql-analytics-endpointeditor — the read-only T-SQL analyst surface over a lakehouse / warehouse / mirror: Synapse serverless SQL over the Delta in ADLS, with views, procs, and security grants. No Microsoft Fabric required.
What it is¶
A SQL analytics endpoint is the read-only T-SQL consumption surface that sits over a lakehouse, warehouse, or mirrored database — the analyst's query layer, exactly like Fabric's auto-provisioned SQL analytics endpoint. In CSA Loom it is Azure-native: the endpoint is Azure Synapse serverless SQL querying the Delta / Parquet in ADLS Gen2 (OPENROWSET / external tables). The editor is the Synapse Studio-style SQL-script surface: an object explorer (views / procs / TVFs / external tables), a Monaco T-SQL editor with catalog IntelliSense, a connect-to-database dropdown, Run / Run-selection, and a Results | Messages pane.
When to use it¶
- Analysts need governed, read-only T-SQL over lake data without provisioning compute.
- You author consumption views / procedures / inline TVFs over OPENROWSET.
- You apply object-level and row-level security so each consumer sees only their slice.
Step-by-step in Loom¶
- Create the item. Choose + New item → SQL analytics endpoint (Data Warehouse). The editor opens at
/items/sql-analytics-endpoint/<id>. - Connect to the endpoint. The endpoint binds to the deployment Synapse serverless SQL pool over the lake's Delta in ADLS. Pick a database in the Connect-to dropdown (master + user databases created via
CREATE DATABASE). - Explore + query. Browse views, stored procedures, table-valued functions, and external tables in the object explorer; write T-SQL in the Monaco editor with catalog-driven IntelliSense and Run (Ctrl+Enter) or Run selection.
- Create consumption objects. Use the New view / New procedure / New function templates to author
CREATE OR ALTER VIEW / PROCEDURE / inline TVFover OPENROWSET (serverless does not support scalar UDFs — the templates emit iTVFs and say so). - Grant access. Apply object-level GRANT / DENY and row-level security (security policies + predicate functions) so analysts get a governed, read-only consumption surface. Export results to CSV / JSON or Open in Excel.
The Azure backend it rides on¶
- Engine: Azure Synapse serverless SQL (
LOOM_SYNAPSE_WORKSPACE); when unset the surface still renders and shows an honest infra-gate. - Data: Delta / Parquet in ADLS Gen2 via OPENROWSET / external tables.
No Fabric required¶
The endpoint is Synapse serverless over ADLS; no Fabric capacity, OneLake, or Power BI workspace is involved.
Learn more¶
- Synapse serverless SQL: https://learn.microsoft.com/azure/synapse-analytics/sql/on-demand-workspace-overview