ETL Pipeline Flowchart: Sources to Data Warehouse

Mermaid code

flowchart LR
    subgraph SRC[Sources]
        S1[(App database)]
        S2[Event logs]
        S3[Partner CSV files]
    end
    S1 --> E[Extract<br/>nightly 02:00]
    S2 --> E
    S3 --> E
    E --> ST[(Staging tables)]
    ST --> V{Schema and<br/>null checks pass?}
    V -->|no| Q[Quarantine table<br/>+ alert]
    V -->|yes| T[Transform<br/>dedupe, join, aggregate]
    T --> W[(Data warehouse)]
    W --> B[BI dashboards]
    W --> M[ML feature store]

Paste this into any tool that renders Mermaid, or into the editor to change the theme and export PNG or SVG. Syntax reference: Flowchart guide.

What this diagram shows

Three sources inside the Sources subgraph feed one Extract step scheduled at 02:00. Everything lands in Staging tables first, then the Schema and null checks pass? diamond either sends the batch to Transform or diverts it to a Quarantine table with an alert. Clean data reaches the Data warehouse, which has two consumers: BI dashboards and an ML feature store. The cylinder shape marks every place data is stored at rest.

How to adapt it

• Rename the three sources to your real systems and add or remove nodes inside the Sources subgraph as needed.
• Change Extract<br/>nightly 02:00 to Stream ingest and replace Staging tables with a Message queue node for a streaming pipeline.
• Split Transform into separate Clean, Join, and Aggregate nodes if each is a separate job with its own owner.
• Add a Data catalog node fed from the warehouse if you publish table documentation.
• For ELT, move Transform after the warehouse node so the diagram shows transformation happening inside the warehouse.

FAQ