Retail project setup guide and references

Hello AdamMichael 

• Mediallion archetecture is recommened and it helps with governance, lineage, and performance. it was suitable for when you have diverrse sources and need structured layeers for analytics.
• Also you can try Star schema or snowflake schema for BI reporting without multiple layers.
• Also you can use Hydread method like use Lakehouse for raw data for Bronze Layer and you can use warehouse for Silver and gold layer
• While coming to deployements MSFT Fabric supports  CI/CD and DevOps style deployments. You can create dev, test, and prod workspaces.
• Then, auutomate promotion of datasets, reports, and warehouse objects across enviironments.
• By using the Git integration for notebooks, dataflows, and code artifacts we can control versions.
• For data retention in one lake,  You can create a  pipelines or notebooks to periodically (in yoru case 3months) delete older data \
• Also, You can organize wareehouse tables by date (monthly partitions concept). Then drop partitions older than 3 months.

Please go through belwo threads, it might helsp you 

Overview of Fabric deployment pipelines - Microsoft Fabric | Microsoft Learn

OneLake Disaster Recovery and Data Protection - Microsoft Fabric | Microsoft Learn

Overview of Fabric deployment pipelines - Microsoft Fabric | Microsoft Learn

Implement medallion lakehouse architecture in Fabric - Microsoft Fabric | Microsoft Learn

DevOps | Fabric Catalyst

Hope this helps you

Thank you!!

Did I answer your question? Mark my post as a solution!

Proud to be a Super User!

Hello AdamMichael 

A recognised pattern of using Fabric is to land ingested data as files into a Lakehouse in /Files area. If you plan to use a code-centric development, you can use Fabric Spark Notebooks to transform the ingested data into "bronze" layer - this transformation does data quality checks, clean data, and apply a structure before landing them into /Tables area inside your Lakehouse. You can apply Medallion architecture to further transform the "bronze" layer into a Fabric Warehouse, into the "silver" layer. This is uaually done using a MERGE INTO SQL construct. A final transformation is then required to aggregate this into the "gold" layer - you may decide to use fact and dimension style modelling on this data.  

The layer that sits on top of gold is your Semantic Model, that includes BI measures and DAX functions. This layer drives your dashboard and reporting. 

So, in summary the chain of transformation is something like this - 

Landing (Lakehouse in /Files) -> Bronze (Lakehouse in /Tables) -> Silver (Warehouse in Silver schema) -> Gold (Warehouse in Gold schema) - > Semantic Model -> Reports and Dashboards

In regards to retention, retention policies in Fabric do not apply to /Files for Lakehouses, so you'd have to apply that yourself, preferably using a Fabric Notebook and by running it as a job. Warehouses automatically cover a 30 days retention by default. The retention window enables Time Travel, Table Clone and Restore operations.