Near Real-Time Data in Fabric That Scales Across Workspaces (Without Materialized Views?)
I’m working through a design pattern in Microsoft Fabric and would love some input from others who have solved something similar.
Scenario:
- Source: Open Mirrored Database (near real-time ingestion)
- Goal: Build near real-time reporting
- Architecture goal: Medallion (Bronze → Silver → Gold)
- Constraint: Needs to scale across workspaces
Proposed Design:
An "admin" workspace which contains a mirrored database.
Multiple workspaces utilizing the medallion architecture:
- Bronze → Lakehouse (shortcuts to tables in mirrored database)
- Silver → Warehouse (cleaned + modeled)
- Gold → Warehouse (serving/reporting layer)
The Problem:
We recently found out that materialized views are not supported in Fabric Warehouses (at least as of now). We're trying to figure out the right pattern to achieve near real-time performance without over-engineering pipelines or introducing latency.
Key Questions:
- How are you achieving near real-time transformation in Warehouses without materialized views?
- Do we need to move Silver/Gold to lake houses to take advantage of materialized views?
- How are you handling cross-workspace data sharing in near real-time scenarios without introducing duplication or sync lag?
- How are you handling data sharing across warehouses in different workspaces?
What I’m Trying to Optimize For:
- Near real-time (<5 min latency preferred)
- Scalable across multiple workspaces
- Simple operational model (fewer moving parts)
Would really appreciate hearing how others are designing around this limitation. Are materialized views just not part of the Warehouse strategy right now, or is there a pattern I’m missing?
Thanks in advance!