Salesforce to Databricks: which connection fits?
Lakeflow Connect, Zero Copy query federation, file sharing or file federation: answer six questions and see which route fits your data, what it needs switched on, and what to watch for.
The routes
Lakeflow Connect Salesforce ingestion (Databricks). It copies Salesforce data into Databricks on a schedule, picking up only what changed. You don't need Data 360. It covers Sales, Service, Platform and most industry clouds, but not Marketing Cloud or B2C Commerce. It needs Unity Catalog and serverless compute. It isn't real-time, formula fields are re-read in full on every run, and hard deletes need a full refresh.
Zero Copy query federation (Databricks reading Data 360). Databricks queries Data 360 where the data sits, over JDBC, and gets live results. It needs a Data 360 license and Unity Catalog. Joins aren't pushed down to Salesforce, so large joins run slowly.
Zero Copy file sharing (Databricks reading Data 360). Databricks reads the files under Data 360 directly and runs the query on its own compute. Databricks says it's the better choice for large amounts of data, with better performance and pushdown than query federation. It needs Unity Catalog and a data share target set up in Salesforce.
Data 360 Zero Copy from Databricks (Salesforce reading Databricks). Data 360 can query Databricks tables live (query federation, with optional caching), or read the files directly (file federation). File federation needs Unity Catalog producing Iceberg metadata, and storage on S3 or Azure Data Lake Storage Gen2 that Salesforce can reach. Salesforce lists the connector for Databricks on AWS and Azure. Models trained in Databricks can also be called from Salesforce.
Databricks' own advice is that the choice depends on how fresh the data must be, how much transformation it needs, and how much of it there is, and that many companies end up using more than one route. This check works the same way.
Sources (checked 1 October 2026)
- Databricks documentation, "Salesforce ingestion connector FAQs" (including "Which Salesforce connector should I use?")
- Databricks documentation, "Salesforce ingestion connector limitations"
- Databricks documentation, "Run federated queries on Salesforce Data 360" (September 2026)
- Databricks documentation, "Lakehouse Federation for Salesforce Data 360 File Sharing" (September 2026)
- Salesforce Developers, "Databricks Connectors" and the Databricks data federation and file federation setup guides
- Salesforce, "Salesforce and Databricks partnership"
Salesforce, Data 360 and Databricks are trademarks of their owners. This page isn't affiliated with or endorsed by either company. Product names and features change often; check the current documentation before you build.
Getting it built
Picking the route is the easy part. Mapping the objects, cleaning the data and keeping the pipeline healthy is the work. MSquare Technology, our Salesforce partner, does data engineering on Databricks and Snowflake as well as Data 360, so it can work on both sides of the connection. MSquare pays Salter Growth when an engagement goes ahead; you pay us nothing.
More: What is data engineering? · Data 360, Databricks or Snowflake? · Salesforce & AI · All free tools
