Pipelines
A pipeline connects one source to one or more targets. Inside it, each entity maps a source table or collection to its destination and defines how that data moves.
Use this page as the map for pipeline work. If you want to build one now, go directly to Build your first pipeline.
Pipeline model
- Source connector
-
Reads changes or snapshot rows from the source system.
- Target connector
-
Writes those records to a destination. A pipeline can have more than one target.
- Entity
-
The replication unit: usually one table or collection and its target mapping.
- Snapshot
-
Copies the current contents of an entity.
- Change data capture (CDC)
-
Continuously forwards inserts, updates, and deletes after replication starts.
The Pipelines screen is the operational home for this model. Select a pipeline to see its entities, health, throughput, and available actions.
Build and configure
| Task | Guide |
|---|---|
Create a pipeline |
|
Add or replace a source or target |
|
Choose tables and collections |
|
Start CDC, run a snapshot, or pause replication |
|
Create groups and ordered chains |
Shape and deliver data
| Goal | Guide |
|---|---|
Filter source rows |
|
Review or change field mappings |
|
Convert, mask, or derive values |
|
Reuse custom logic across entities or compose one column from several inputs |
|
Run custom transformation code |
|
Control target write behavior |
|
Create missing target tables |
Operate a pipeline
-
Read Pipeline states and error recovery before responding to an unhealthy pipeline.
-
Use Snapshot tasks for large initial loads and interrupted snapshots.
-
Put a pipeline in maintenance mode before configuration changes that require agents to disconnect.
-
Configure monitoring and alerts for production.