PROVIDER Β· Microsoft unified analytics service
Azure Synapse Analytics
Azure Synapse Analytics consolidates dedicated and serverless SQL, Spark pools, and integration pipelines into one workspace with native Power BI, Entra, and Purview ties as Microsoft's enterprise analytics platform on Azure.
4pts
Pros
Cons
Product Information
Detailed Notes
Overview
Azure Synapse Analytics launched in 2020 out of SQL Data Warehouse, consolidating a dedicated SQL pool, serverless SQL, Spark pools, and integration pipelines into one workspace as Microsoft's enterprise analytics platform on Azure.
For Backend & Database work the value is deep integration with the Azure ecosystem and Power BI, since data lives in ADLS Gen2, Spark and SQL engines share metadata within one workspace, Power BI connects directly, and existing Microsoft identity plus security policies carry over.
Practical concerns are roadmap and cost, because Microsoft now pushes Microsoft Fabric and some Synapse capabilities such as Data Factory and real-time analytics are migrating there, so new projects should confirm the long-term path, and dedicated SQL pools bill on DWUs even when idle, making pause policies mandatory.
Core Strengths
- Dedicated SQL pools deliver petabyte-scale MPP warehousing with columnstore and result-set caching
- Serverless SQL pools query lake files billed by scanned volume without provisioning
- Spark pools and SQL engines share metadata for engineering and analytics in one workspace
- Native integration with Power BI, Purview, and Microsoft Entra
Product Ecosystem
Dedicated SQL pool
A DWU-provisioned MPP warehouse for steady high-frequency enterprise reporting.
Serverless SQL pool
Byte-scanned billing for lake exploration and light transformation without provisioning.
Spark pools and pipelines
Spark handles engineering and machine learning while Synapse Pipelines orchestrate and schedule.
Drawbacks
- The roadmap shifts toward Microsoft Fabric, so confirm official plans before committing
- Dedicated pools bill on DWUs and misconfigured pause policies create idle cost
- Many components and concepts raise onboarding cost above a plain warehouse
- Azure-bound and not portable to other clouds or domestic stacks
Use Cases
- Enterprise reporting on Azure (β β β β β ) - Direct Power BI connection
- Lake exploration and light transformation (β β β β ) - Serverless SQL on demand
- Unified analytics in Microsoft stacks (β β β β ) - Entra and Purview integration
- Greenfield analytics platforms (β β β ) - Evaluate the Fabric roadmap first
Pricing
| Plan | Price | Scope |
|---|---|---|
| Serverless SQL | About USD 5 per TB scanned | No provisioning for lake queries |
| Dedicated SQL pool | Per DWU-hour | Pausable, with idle cost if forgotten |
| Spark pools | Per vCore and node-hour | Created on demand with autoscale |
Synapse cost management centres on pausing dedicated pools, since DWU bills hourly and leaving them up overnight wastes money, while exploration is cheaper on serverless and Spark pools bill per node-minute with autoscale and short idle timeouts.
FAQ
-
Synapse or Fabric? Continue Synapse where investment exists while planning migration, and evaluate Microsoft Fabric first for greenfield work since Microsoft centres unified analytics there.
-
Is serverless SQL production ready? It suits lake exploration and light view layers rather than high-concurrency heavy reporting, which belongs on dedicated pools or a Fabric warehouse.
-
How is cost controlled? Auto-pause dedicated pools, bound serverless scan volume, configure autoscale with short idle timeouts for Spark, and enable cost management budget alerts.