Snowflake track
Snowflake Data Platform with dbt and Airflow
You will take WideWorldImporters, Microsoft's sample wholesale database on SQL Server, and build a complete analytics platform on Snowflake: a raw Bronze layer loaded incrementally, a Silver layer and Gold star schemas built with dbt, a Sales data mart for reporting, tests and a reconciliation that proves every number, all run by one simple Airflow DAG.
Start with the design
Chapter 0 shows the five design diagrams at full size, before any code. Keep it open in a second tab: every later chapter zooms into one part of them with a smaller diagram.
What you will have at the end
| Layer | Where | What it holds |
|---|---|---|
| Source | SQL Server 2022 · WideWorldImporters | 31 OLTP tables in 4 schemas |
| Bronze | Snowflake · WWI_RAW.BRONZE | 31 raw tables, loaded incrementally, with batch and load metadata |
| Silver | WWI_ANALYTICS.SILVER | 31 staging views, 5 SCD Type 2 snapshots, 4 intermediate models |
| Gold | WWI_ANALYTICS.GOLD | 8 conformed dimensions and 6 facts |
| Data mart | WWI_ANALYTICS.MART_SALES | 1 fact (one row per invoice line) and 6 dimensions |
| Orchestration | Airflow on the lab VM | 1 DAG, 10 BashOperator tasks, every night at 02:00 |