When and how to migrate to Azure Synapse?
summary
Migrating an on‑prem data warehouse to Azure Synapse means rethinking architecture: Synapse uses 60 distributions and MPP compute, so table design, distribution choice and partitioning are decisive — clustered columnstore needs roughly 1M rows per distribution (≈60M rows total) and replicated tables suit <2GB. Use PolyBase (or ADF orchestration) to load files from Blob/ADLS; demo imports 77M rows in 1m16s from 13 files. Expect DWU scale‑up during loads, monitor DMVs, avoid OLTP patterns, and prefer hash for large facts, replicated for small dimensions.
AI-written summary · about
chapters (23)
- 0:00Introduction
- 0:31Overview
- 0:56Kamils background
- 3:06Azure Synapse
- 4:20Modern Data Warehouse
- 6:45Azure Synapse vs SQL Data Warehouse
- 7:33Azure Synapse Architecture
- 9:44Analytics Runtime
- 10:15SQL Provisioning
- 10:51Data Distribution
- 14:27Data Distribution Options
- 15:03Visualizing Data Distribution Options
- 16:25What workloads are suitable
- 17:02What workloads are not suitable
- 18:25Demo
- 19:38Launch Synapse Studio
- 20:00Develop App
- 21:36External File Format
- 27:31Hash Distribution
- 29:20Hash Distribution Exercise
- 31:57Replication Distribution Exercise
- 39:30How to prepare
- 43:22Importing data to Azure Synapse
related talks
54mLoading Data into Azure SQL DW (Synapse Analytics)
12mModern Data Warehouse for Small and Medium Business
46mThe Ultimate Jigsaw - Architecting Azure Synapse Analytics
19mNever ETL again, thanks to Synapse Link. Really?
50mPower BI + Azure Synapse Analytics: Beginning your journey
51mSupercharge Power BI with Azure Synapse Analytics- 19mNever ETL again, thanks to Synapse Link. Really?
45mWhat's New and Cool in Azure Synapse Analytics