examples/data-warehouse.md

GitHub

Data Warehouse

Star-schema data warehouse with Redshift, Glue ETL, and QuickSight dashboards.

Key Elements

Component Stencil
Redshift mxgraph.aws4.redshift
Redshift RA3 mxgraph.aws4.redshift_ra3
Glue mxgraph.aws4.glue
Glue Data Catalog mxgraph.aws4.glue_data_catalog
QuickSight mxgraph.aws4.quicksight
S3 mxgraph.aws4.s3
RDS mxgraph.aws4.rds

Example

Enterprise data warehouse with multi-source ETL and executive dashboards:

@startuml
left to right direction

rectangle "Operational Sources" {
  mxgraph.aws4.rds "Orders DB\n(RDS)" as orders
  mxgraph.aws4.rds "CRM DB\n(RDS)" as crm
  mxgraph.aws4.s3 "Marketing\nCSVs (S3)" as csvs
}

rectangle "ETL Layer" {
  mxgraph.aws4.glue "Glue\nETL" as glue
  mxgraph.aws4.glue_data_catalog "Data\nCatalog" as catalog
  mxgraph.aws4.s3 "Staging\n(S3)" as staging
}

rectangle "Data Warehouse" {
  mxgraph.aws4.redshift "Redshift\nCluster" as rs
  mxgraph.aws4.redshift_ra3 "RA3\nNodes" as nodes
}

rectangle "BI Layer" {
  mxgraph.aws4.quicksight "Executive\nDashboard" as dash
  mxgraph.aws4.quicksight "Sales\nReports" as sales
}

orders --> glue
crm --> glue
csvs --> glue
glue --> catalog
glue --> staging
staging --> rs
rs --> nodes
rs --> dash
rs --> sales
@enduml

Pattern Notes

  1. Glue ETL: extracts from RDS/S3 sources, transforms to star schema (fact + dimension tables), loads to Redshift
  2. Data Catalog: metadata registry — tables, schemas, and data lineage for governance
  3. Staging in S3: intermediate Parquet files before COPY into Redshift for fault tolerance
  4. Multiple dashboards: QuickSight SPICE datasets connected to Redshift for sub-second queries
Icon Stencil Use When
Redshift ML mxgraph.aws4.redshift_ml In-warehouse ML inference
Streaming Ingestion mxgraph.aws4.redshift_streaming_ingestion Real-time Kinesis ingestion
Athena mxgraph.aws4.athena Serverless SQL on S3 fallback
Lake Formation mxgraph.aws4.lake_formation Cross-account governance
Data Exchange mxgraph.aws4.data_exchange Third-party data sets
EMR mxgraph.aws4.emr Heavy ETL before warehouse load