Lakeflow Connect versus Snowflake Openflow for ingestion

Summary

• Lakeflow Connect is the managed ingestion layer of Databricks Lakeflow: connectors for databases, SaaS applications, files and message buses run as ingestion pipelines that write streaming tables governed by Unity Catalog.
• Snowflake Openflow is an integration service built on Apache NiFi. Its connectors are curated, versioned NiFi flow definitions, and you can build your own dataflows from Snowflake and NiFi processors and controller services.
• Openflow has two deployment types: Snowflake Deployments, which run on Snowpark Container Services, and Bring Your Own Cloud (BYOC).
• Both cover change data capture from MySQL, PostgreSQL, Oracle and SQL Server. In Lakeflow Connect the stage varies by connector: SQL Server is GA, PostgreSQL and MySQL are in Public Preview, and Oracle is in Beta.
• The destination usually decides it: Lakeflow Connect lands data in Unity Catalog for the Databricks lakehouse, and Openflow's documented use cases bring data into Snowflake.

Lakeflow Connect versus Snowflake Openflow for ingestion

Lakeflow Connect and Snowflake Openflow are each platform's managed ingestion service, and they're built on different models. Lakeflow Connect is configuration-driven: you choose a managed connector, store credentials in a Unity Catalog connection, and the connector runs an ingestion pipeline that writes streaming tables. Openflow is flow-driven: it's built on Apache NiFi, ships connectors as versioned NiFi flow definitions, and lets you assemble your own flows from processors. We believe Lakeflow Connect is the right choice when the destination is the Databricks lakehouse, because ingestion, governance and orchestration then sit in one platform.

What is Snowflake Openflow?

Snowflake describes Openflow as "an integration service that connects any data source and any destination", built on Apache NiFi. Data flows execute in runtimes inside a deployment, and a control plane manages them through the Openflow canvas and APIs. Snowflake Deployments run runtimes as Snowpark Container Services services on a compute pool. BYOC deployments run the data plane in your own cloud environment while Snowflake manages the service and control plane. The FAQ below covers availability.

What is Lakeflow Connect?

Lakeflow Connect organizes connectors by source type: database (CDC), query-based, SaaS, file and streaming connectors, plus community and custom connectors. Database connectors add an ingestion gateway. The first run ingests all selected data, and later runs ingest only what changed, when the source allows it.

Lakeflow Connect vs Snowflake Openflow: how they compare

Capability	Databricks Lakeflow Connect	Snowflake Openflow
Building block	Managed connectors, configured as ingestion pipelines	Connectors are curated, versioned Apache NiFi flow definitions
Custom pipelines	SQL and Spark APIs: Auto Loader, Spark Declarative Pipelines (SDP) and Structured Streaming, plus community and custom connectors (Beta)	Custom dataflows built from Snowflake and NiFi processors and controller services
Where it runs	SaaS and streaming connectors run ingestion pipelines on serverless compute; database connectors add an ingestion gateway	Snowflake Deployments on Snowpark Container Services, or BYOC in your own cloud
Database CDC	MySQL, PostgreSQL and SQL Server through CDC, and Oracle through LogMiner	Connectors for MySQL, PostgreSQL, Oracle and MongoDB, plus SQL Server through Change Tracking or CDC
SaaS sources	Each managed connector handles authentication, incremental reads, schema evolution and automated retries	Connectors for SaaS sources such as HubSpot, Jira Cloud and Shopify
Files and documents	Managed connectors for Google Drive and SharePoint (GA) and OneDrive (Beta); Auto Loader for cloud object storage	Connectors for Google Drive, Box and SharePoint, with content prepared for chat with Snowflake Cortex
Event streams	Managed Kafka and RabbitMQ connectors (Beta), and Zerobus Ingest for push-based writes	Connectors for Apache Kafka and Amazon Kinesis Data Streams
Destination	Streaming tables, which are Delta tables, governed by Unity Catalog	Snowflake tables, with opt-in Iceberg table destinations (Preview) for the MySQL, PostgreSQL and Oracle connectors
Scheduling	Each schedule creates a Lakeflow Jobs job with the pipeline as a task	Runtimes host the pipelines, managed through the Openflow canvas or APIs
Credentials	A Unity Catalog connection stores the source's authentication details	Snowflake Managed Token is the default runtime authentication, with Snowflake secrets for credentials
Deployment as code	Declarative Automation Bundles manage pipelines across environments	Gen 2 adds SQL-first management of deployments, runtimes and connectors; gen 2 connector configuration is in Public Preview

How Databricks approaches ingestion

Databricks recommends starting with the most managed layer. Managed connectors in Lakeflow Connect come first. Next is SDP, which adds a declarative framework on top of Structured Streaming, and then Structured Streaming itself for full control. Files that land in cloud storage go through Auto Loader, which processes new files incrementally and keeps exactly-once guarantees through its checkpoint. Producers that push records can use Zerobus Ingest, a push-based API that writes straight into Unity Catalog Delta tables with no message bus to run. Databricks announced its general availability on February 23, 2026.
Lakeflow Connect provides event logs, pipeline health metrics and data quality metrics, and a failed connector retries with exponential backoff.
Check each connector's stage before you plan around it. SQL Server connector is GA, the PostgreSQL and MySQL connectors are in Public Preview (request access through your Databricks account team), and the Oracle connector is in Beta.

FAQs

Is Snowflake Openflow generally available?

Snowflake's documentation states that Openflow Snowflake Deployments are available to all accounts in AWS, Azure and GCP commercial regions, and BYOC deployments to all accounts in AWS commercial regions. It also notes that Openflow isn't automatically available in trial accounts and that you contact your Snowflake account team to enable it.

Can Lakeflow Connect ingest from Apache Kafka?

Yes. The managed Kafka connector in Lakeflow Connect is in Beta and writes Kafka topics into streaming tables. You can also read Kafka with SDP or Structured Streaming for finer control.

What if there's no managed connector for my source?

In Databricks, you can use SQL and Spark APIs, an open-source community connector, or a custom connector that runs in your own workspace (both are in Beta). In Openflow, you build a dataflow from processors and controller services.