Why it matters

Data Hopper EDW turns Apache Hop into a warehouse factory: you describe sources, vaults, and marts as model files and Hop generates and runs the pipelines that build them. Without a map, six model file types and four layers look like six products. With the map, every later module is one more step on the same road.

This module gives you that map and shows where each job happens in Hop GUI. No database is needed: you only read the retail-example project. Budget 25 minutes.

Concepts

Four layers, one road

Source systems land in a catalog, the catalog feeds a raw Data Vault, that feeds a Business Vault, and that feeds dimensional marts. Each layer has one job: sources are contracts, the raw vault is history, the business vault is interpretation, the marts are consumption.

The four layers of a Data Hopper EDW

The catalog is the contract bus

Nothing is wired by JDBC URL. A source model publishes feeds into the Data Catalog under hop/{project}/sources, hubs and satellites bind to those feed names, and a load publishes the target layouts back. Rename a column in the catalog and every model that uses it knows.

The Data Catalog is the contract bus

The resource definition group is the spine

A resource definition group is the list of warehouse models one warehouse consists of. Harvest, schema validation, version tags, and loading all take the group as their scope — which is why you create the group early and grow it, instead of starting a new one per subject area.

Model files

Extension What it describes

.hsm

Source model: tables, primary/foreign keys, joins, JSON and pipeline feeds

.hdv

Raw Data Vault model: hubs, links, satellites, reference tables

.hbv

Business Vault model: SCD2, PIT, bridges, SQL business tables

.hdm

Dimensional model: facts, dimensions, date dimensions

.hsl

Semantic layer over a .hdm (consumption metadata, saved selections)

.hem

Execution map: what ran, in what order, with what result

.hlv

Hop Lineage View: a graph definition over a lineage backend

Build order and the four controls

Build in the order the catalog allows: source model → publish to catalog → generate the raw vault → create the resource definition group → first load → Business Vault → dimensional marts. Around that run four controls: Check model (build time), metadata harvest, the schema gate, and data quality. Confusing those four is the most common beginner mistake — they answer different questions.

The four controls and the questions they answer

Where you work in Hop GUI

The EDW Journey perspective draws a resource definition group as the warehouse map: sources, controls, raw vault, business vault, dimensional, orchestration, outputs. It is the fastest way to see what a warehouse contains and which piece owns a table.

The EDW Journey tree

Guided walkthrough

Open the finished warehouse

  1. Start Hop GUI and open the project folder retail-example/ (File → Open project).

  2. Open the EDW Journey perspective (perspectives row, after Data catalog).

  3. In the group combo at the top, select retail-sources.

  4. Expand 1. Sources: you see the source model .hsm files discovered from the catalog, then the catalog feeds themselves (namespace hop/retail-example/sources).

  5. Expand 3. Raw Data Vault and its child retail-360.hdv: the hub, link, and satellite table names appear under it.

  6. Select retail-360.hdv and press Documentation — the offline HTML guide for that model opens. This is the same documentation set these training modules ship in (toolbar EDW button, or Help → EDW documentation), and it works with no network.

  7. Expand 4. Business Vault, 5. Dimensional, 7. Orchestration, and 8. Outputs to see the rest of the same warehouse: .hbv and .hdm files, the workflows that name the group, and where reports and execution maps land.

    The resource definition group the journey tree is built from
  8. Select 1. Sources and press Details → Open catalog to see the published feeds that the whole warehouse is wired to.

  9. Open models/retail-360.hdv in the modeler and use Browse on one hub: the canvas centers the table and its neighbours, which is the fastest way to read a vault you did not write.

    The retail-360 Data Vault model

Try it yourself

Pick one business concept that matters to you (customer, order, product) and write down its path through the warehouse, from the journey tree only:

  1. Find the hub for that concept under 3. Raw Data Vault.

  2. Find at least one satellite of that hub and note the descriptive columns it would carry — open the table dialog (Details → Open model, then double-click the table) and read the Fields.

  3. Find the catalog feed the satellite’s hub binds to under 1. Sources.

Check your work: the hub name starts with hub_, the satellite name with sat_, and the feed name is a catalog entry — not a database table. If you wrote a test_edw table name, re-read the Concepts section on the catalog above.

Questionnaire

Answer the six questions. Every answer comes with its reasoning, and Copy my results gives you a text summary to keep.

1. Which file extension holds the raw Data Vault model?

2. Where does a hub get the name of the source it reads?

3. What is a resource definition group?

4. In the recommended build order, when is the raw Data Vault generated?

5. Which statement about the four controls is correct?

6. Scenario: your team wants to add a Business Vault customer view on top of the existing warehouse. What is the right next step?

Where to go next