All pages
❄ Snowflake | Openflow SQL Server Connector
Source Microsoft SQL Server
Destination Snowflake
Method Change Tracking or Change Data Capture

SQL Server, replicated into Snowflake.

The Openflow SQL Server connector streams Microsoft SQL Server into Snowflake in near real time. It is Snowflake-native, with no third-party tooling and no hand-built CDC pipelines. One connector reads every database on the instance and lands each into its own schema.

Microsoft SQL Server
Source instance · Change Tracking or CDC enabled · many databases
→TDS / TCP-IP
Openflow runtime
SPCS / NiFi · snapshot, then incremental merge
→load
Snowflake
Destination + journal tables · one schema per source DB

Where to start

Start here
Evaluate
What the connector does, the pain it removes, and how it compares to Fivetran and hand-built ETL. Then the best-fit scenarios: warehouse sync, medallion bronze, ETL cost takeout, audit and history.
Overview2 sections
Open →
Decide
Choose a mode
The two decisions to make before you configure anything: Change Tracking versus Change Data Capture, and one connector per database versus one for the whole instance. Each with a side-by-side comparison and a recommendation.
Decision guide2 sections
Compare →
Configure
Network
The protocol stack, port requirements and TDS encryption modes, the External Access Integration setup with the exact SQL, and the errors you are most likely to hit with the check for each one.
Setup3 sections
Configure →
Operate
How the pipeline runs
Snapshot load then incremental merge, what happens in each phase, and the levers that matter when either one gets slow or expensive. Includes stream staleness, which is a hard failure boundary rather than a tuning problem.
Runbook5 sections
Read →
Look up
Processor reference
Every processor in the connector flow, grouped by responsibility, for snapshot, incremental and staleness prevention. Dense tables, meant for looking something up rather than reading start to finish.
Reference3 sections
Look up →

What it replaces

Replicating SQL Server into Snowflake has usually meant one of three things: paying per row for a third-party CDC vendor, hand-building a pipeline that reads change tables and merges them yourself, or accepting a nightly batch and the staleness that comes with it. The connector is a managed alternative that runs inside Snowflake, reads the change feed SQL Server already produces, and keeps the destination current without a pipeline for you to maintain.

Learn more
Openflow Connector setup Change Tracking vs CDC SQL Server network configuration TDS 8 Reference flow definitions

The processor reference is drawn from the flow definitions in the repository above, which hold the connector configuration used throughout this document.