Get started

Set up a single replicate pipeline using the public Northwind OData V4 API: remote Products → local table. Swap the URL and model for your own API when ready.

Prerequisites

• Node.js >= 22
• @sap/cds >= 9 (CDS 9 and CDS 10 both supported)
• Network access to https://services.odata.org

1. Install the plugin

npm add cds-data-pipeline

The plugin auto-registers the pipeline engine — no extra config needed for addPipeline. Connect with cds.connect.to('data-pipeline').

AI assistants

Coming soon. Agent guidance and task skills for coding assistants will ship in a future release.

Option A — reuse management API from the plugin (no project CDS files):

{
  "cds": {
    "requires": {
      "data-pipeline": {
        "impl": "cds-data-pipeline",
        "management": { "reuse": { "api": true } }
      }
    }
  }
}

Option B — manual CDS imports (explicit control):

Tracker schema in db/schema.cds:

using from 'cds-data-pipeline/db';

Management OData in srv/pipeline-mgmt.cds:

using from 'cds-data-pipeline/index.cds';

Or import tracker and service separately:

using from 'cds-data-pipeline/db';
using from 'cds-data-pipeline/srv/DataPipelineManagementService';

The plugin ships no auth annotations — add your own as needed. See Management Service.

2. Import the Northwind OData API

The plugin needs a connected service name and entity names from your imported model — the standard capire OData import flow. Skip the expander if you already have an imported service.

cds import and Northwind (step-by-step)

1. Download metadata (example: Northwind V4 — save as .edmx):

curl -sL 'https://services.odata.org/V4/Northwind/Northwind.svc/$metadata' -o northwind.edmx

2. Import from your CAP project root (adjust the path if the file lives elsewhere):

cds import northwind.edmx

3. Connectivity — set credentials.url to the service root (no $metadata suffix), e.g. https://services.odata.org/V4/Northwind/Northwind.svc, or use a BTP destination.

After import, CAP prints a using hint:

using { northwind as external } from './external/northwind';

The service name (here northwind) is what you pass as source.service in addPipeline.

3. Define a consumption view

Model the local target as a projection on northwind.Products. This example restricts columns, renames one field (ProductName → ProductTitle), and excludes discontinued rows:

using { northwind } from '../srv/external/northwind';

@cds.persistence.table
entity LocalProducts as projection on northwind.Products {
    ProductID,
    ProductName as ProductTitle,
    UnitPrice,
    UnitsInStock,
} where Discontinued = false;

Place this in your app namespace (e.g. my.app in db/schema.cds) so the fully qualified name is my.app.LocalProducts. See Concepts → Consumption views for details.

4. Register your first pipeline

In server.js, connect after CAP has served and call addPipeline. The engine infers column mappings and renames from the consumption view. Northwind Products has no change-timestamp field, so we use mode: 'full' here (for delta, see Built-in replicate).

const cds = require('@sap/cds');

cds.on('served', async () => {
    const pipelines = await cds.connect.to('data-pipeline');

    await pipelines.addPipeline({
        name: 'NorthwindProducts',
        source: { service: 'northwind', entity: 'Products' },
        target: { entity: 'my.app.LocalProducts' },
        mode: 'full',
        schedule: 600_000, // optional: every 10 minutes; omit and use execute (below)
    });

    // Example: tweak one column after the default mapper (do not use `on` here — it would replace the default)
    pipelines.after('PIPELINE.MAP', 'NorthwindProducts', (_results, req) => {
        for (const row of req.data.targetRecords) {
            if (row.ProductTitle != null) {
                row.ProductTitle = String(row.ProductTitle).trim().toUpperCase();
            }
        }
    });
});

module.exports = cds.server;

after('PIPELINE.MAP') runs per batch on top of the built-in mapping. req.data.targetRecords already uses local names (e.g. ProductTitle, not ProductName). Use before / after to layer behavior — on replaces the default mapper entirely. See Event hooks.

5. Deploy and run

• SQLite / local: cds deploy (or your usual profile), then cds watch / cds serve.
• HANA: include the plugin DB model in production build and deploy with your HDI flow.

6. Open the Pipeline Console

Enable reuse in package.json:

{
  "cds": {
    "requires": {
      "data-pipeline": {
        "impl": "cds-data-pipeline",
        "management": {
          "reuse": {
            "api": true,
            "console": true
          }
        }
      }
    }
  }
}

cds deploy
cds watch
# → http://localhost:4004/pipeline-console/index.html

BTP and project-owned UI: cds add pipeline-console. Full decision tree: Feature activation and Pipeline Console.

Without a UI, use the management OData API directly (GET /pipeline/Pipelines, GET /pipeline/PipelineRuns). See Management Service.

7. Query the service and check the data

1. Trigger a run (if no schedule set): POST /pipeline/execute with body { "name": "NorthwindProducts", "mode": "full" }. See execute.
2. Check status: GET /pipeline/Pipelines(‘NorthwindProducts’) or GET /pipeline/status(name=’NorthwindProducts’).
3. Query data: access my.app.LocalProducts via your app service or CDS and confirm rows match non-discontinued Northwind products.

Next

• Pipeline Console — mount options, mashups, security
• Recipes — replicate, materialize, fan-in, custom adapters, hooks, scheduling
• Concepts — vocabulary, inference, consumption views
• Feature catalog