| ID | product.ingestion |
|---|---|
| Description | Collects data from operational sources via CDC, batch APIs and event streams. |
| Key | ingestion |
| Type | system |
| Team | Data Ingestion |
| Owner | ingestion@dataplatform.example |
| Status | active |
| Technologies | Java, Kafka Connect, Debezium |
| Attributes | latency: near-real-time, paths: cdc,batch,stream |
| Tags |
Ingestion is the front door of the data platform. Every row change in a production database, every SaaS API export and every domain event on the bus passes through here before it can be stored, transformed or queried. If data enters the lakehouse at all, it enters through Ingestion first.
Responsibility in one sentence
Capture operational data from databases, APIs and event streams — with resumable offsets, schema awareness and deduplication — and land it in the bronze layer of the lakehouse.
Could not build diagram for model "example-data-processing".
Operational DB changes replicate to the lake in near real time:
Domain events from microservices enter the same path through the stream router:
Schema changes
When upstream DDL arrives, the Schema Evolver classifies the change — see the schema-change scenario in specs.
4 outgoing links
| Target object | Label | Link type | Relation | Raw id | Required | Description | |
|---|---|---|---|---|---|---|---|
| reads CDC from | — | links | external_systems.production_dbs | — | — | ||
| polls | — | links | external_systems.saas_apis | — | — | ||
| subscribes to | — | links | external_systems.event_bus | — | — | ||
| lands data in | — | links | product.storage | — | — |