Pattern - Gold Materialized Views

Description

Suitable for gold-layer serving tables built from silver (or other upstream) tables using materialized views.

Use when:

  • You want batch or incremental batch serving at gold (dimensions, facts, aggregates).

  • Lower latency than a scheduled notebook job is sufficient.

  • You prefer declarative SQL over streaming-first gold pipelines.

Layers: Gold

Models: Dimensional (dimensions and facts), data marts, curated aggregates

Note

For gold workloads, materialized views are generally the first choice unless you need streaming-first gold with lower latency. See Materialized Views.

Data Flow Components:






No.


Component


Description


M / O






1


Upstream table(s)


Silver (or gold) tables referenced in MV SQL — often current-state SCD2 or conformed entities.


M




2


Materialized view spec


dataFlowType: materialized_view with one or more MV definitions (including chained MVs).


M











Feature Support





Supported


Not Supported






    

  • Full and incremental refresh policies
    • 

  • Chained materialized views
    • 

  • Joins across upstream tables in MV SQL
    • 




    

  • Multi-source streaming topology (use silver streaming patterns first)
    • 













Sample


    

  • Bundle: samples/pattern-samples
    • 

  • Pipeline: Lakeflow Framework - Pattern - Base Gold Samples Pipeline
    • 

  • Sample: samples/pattern-samples/src/dataflows/base_samples/gold/dataflowspec/gold_materialized_views_main.json
    • 



For stream-static gold dimensions (streaming silver into gold), see samples/pattern-samples/src/dataflows/stream_static_samples/dataflowspec/gold_dim_customer_json_main.json and the stream-static pattern pages.