DataEngBytes 2023 - BRI-T1-04 - Mimoune Djouallah
summary
Early-access report on Microsoft Fabric claims a unified data stack that actually reads and writes Delta Lake files, links SQL warehouse, Spark and Power BI without copy steps, and exposes direct lake mode where Power BI loads remote delta into RAM and behaves like import. Fabric shows bursting (serverless extra cores) and an F64 tier that gives free Power BI reader. Tests run TPCH SF100 (600M rows, 22 queries) in ~90s; pipeline demo pulls Australian electricity prices every five minutes. Still preview—utilization metrics are missing so cost/perf for production remains uncertain.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 3:16Database innovation and Fabric origins
- 7:14Fabric architecture vision
- 10:46Fabric pricing models
- 13:57Serverless compute costs
- 16:29Delta table format
- 19:19Fabric Data Warehouse performance
- 22:09Power BI Direct Lake mode
- 25:07Demo with electricity data
- 27:37Fabric adoption considerations
- 29:04Final thoughts
related talks
47mData analytics for the era of AI
21mDataEngBytes 2023 - SYD-T1-06 - Sana Sanai
51mA Deep Dive into Microsoft Fabric Data Warehouse
46mMicrosoft Fabric Synapse data warehouse Q&A
50mFabric: What's new and what's next
50mFabric Direct Lake Deep Dive
50mZero to Lakehouse in Microsoft Fabric
50mMicrosoft Fabric Data Factory Q&A
