In private development
Planned, with no date. SQL Server is a source only: Emolysis reads from it and never deploys to it.
SQL Server
SQL Server as a source.
SQL Server is a planned source, not built yet. As designed, Emolysis reads from it and never deploys to it: tables and change data land on your production platform through tested Ingest (batch), Incremental append and CDC merge pipelines.
Planned
Connect
SQL authentication or Microsoft Entra ID
Sample
A slice into DuckDB on your machine
Prove
Golden tests on DuckDB
Land
Bronze tables on your platform
Overview
SQL Server as a source.
SQL Server as a source is planned, with no date yet. As designed, Emolysis reads from SQL Server and never deploys to it. Data from SQL Server lands on your production platform: Databricks first, then Snowflake and Microsoft Fabric.
Sources use the same profile format as platforms, so the agent knows how to read them correctly: T-SQL with bracket-quoted identifiers, the types that need care, and the change options SQL Server offers.
Three pipeline shapes do the work. Ingest (batch) lands full extracts in bronze. Incremental append reads only new rows by a watermark. CDC merge applies inserts, updates and deletes from Change Data Capture or Change Tracking. Which change option comes first is not decided.
For development, Emolysis samples a slice of the source into DuckDB on your machine, so the golden tests run locally. The deployed pipeline runs on your platform, under your account, like every other pipeline.
Source profile
What the profile captures.
Sources use the same profile format as platforms, with fields of their own. This is what the SQL Server profile is drafted to capture.
- Dialect
- T-SQL, including bracket-quoted identifiers
- Types that need care
- datetime versus datetime2, money, nvarchar, uniqueidentifier and rowversion
- Change options
- Change Data Capture, Change Tracking, and rowversion or updated-at watermarks
- Sign-in
- SQL authentication or Microsoft Entra ID
- Sampling
- A slice of a table into DuckDB on your machine
- Deploy target
- None. Data lands on Databricks, Snowflake or Microsoft Fabric
Add a source
Adding a source.
How adding SQL Server to a project works, as designed, step by step. None of it is built yet.
Add a source
In a project, you add SQL Server as a source and enter its connection details.
Sign in
With SQL authentication or Microsoft Entra ID. Credentials are kept in your machine's credential store and never sent to Emolysis.
Sample a slice
For development, Emolysis samples a slice of a table into DuckDB on your machine, so golden tests run against the real shape of your data.
Find the changes
Each table needs a way to find changes: Change Data Capture, Change Tracking, or a rowversion or updated-at watermark. Which of these comes first is not decided.
Landing
Lands on your platform.
SQL Server has no deploy target in Emolysis. Its data feeds the Ingest (batch), Incremental append and CDC merge shapes, which deploy to your production platform and run there.
Types mapped on purpose
The profile names the types that need care, such as datetime2, money and uniqueidentifier, so mapping them to your platform's types is explicit, not guessed.
Bracket-quoted names
T-SQL quotes identifiers with brackets. Names with spaces or reserved words are read correctly, then normalized in silver by Dedupe / conform.
Changes applied in order
CDC merge applies only the latest change for each key in a batch, in order, so updates are not applied out of order and deletes are not ignored.
Bronze first
Every load lands in bronze tables on your platform, with load metadata, before Dedupe / conform builds silver tables from it.
Shapes
Three shapes read from SQL Server.
SQL Server feeds three shapes. After landing, the rest of the pipeline runs on your platform like any other.
- Ingest (batch)Full extracts of SQL Server tables land in bronze on your platform, typed on load, with types such as datetime2, money and uniqueidentifier mapped deliberately.
- Incremental appendA rowversion or updated-at watermark selects the rows added since the last run. The watermark moves forward only after the load commits.
- CDC mergeChange Data Capture or Change Tracking provides the change feed. The merge applies the latest change for each key, in order, including deletes.
All ten shapes, with their templates and golden tests, are on the pipeline shapes page.
Read next
Glossary
Terms on this page
Questions
SQL Server questions.
Answers about SQL Server and how it fits. More in the FAQ.
Can Emolysis deploy to SQL Server?
No. SQL Server is planned as a source only. Emolysis reads from it and lands the data on your production platform: Databricks first, then Snowflake and Microsoft Fabric. If you need computed values back in an operational database, the Reverse ETL shape does that.
Which SQL Server change option does Emolysis use?
Change Data Capture, Change Tracking, and rowversion or updated-at watermarks are all in the source profile. Which of these comes first is not decided. Change Data Capture and Change Tracking feed the CDC merge shape; watermarks feed Incremental append.
Do production loads run on my laptop?
No. Sampling a slice into DuckDB is for development and golden tests. What you deploy runs on your production platform, under your account, like every other pipeline Emolysis deploys.
Where are SQL Server credentials kept?
In your machine's credential store, the same as platform credentials. They are never sent to Emolysis.
Is there a date for SQL Server as a source?
There is no date yet. Support is announced on the news page, together with the shapes whose templates pass their tests for it.
Nothing is released yet
Tell us what you load.
SQL Server as a source is planned, with no date yet. Early access starts with Databricks teams. If you load from SQL Server, tell us which change option you rely on.