It is a data warehouse and a unified analytics platform
It uses ADLS Gen 2 (Data Lake) as its underlying datastore
It is a PaaS (Platform as a Service) offering
Data Warehouse Unit (DWU): Unit used to represent the cost/ compute of SQL Pool
Unpredictable Workflows: Serverless SQL
Predictable Workflows: Dedicated SQL
Key Features
Allows to build ETL/ ELT pipelines
Ingest data from more than 95 connectors
Deep integration with Apache Spark
Use T-SQL statements to query data
Integrated with BI and AI Tools
Synapse can connect to Apache Spark
The SparkML and AzureML model integration features are built-in
Has support for Delta File Format
Allows to write Spark in .NET as well
Synapse SQL
The distributed version of T-SQL
Supports streaming features that allow saving data into SQL Tables
Allows to integrate AI with SQL
Serverless Pool
In Serverless the topology changes by adding, removing or failovers of nodes
A Serverless Pool is automatically created when a new Synapse Workspace is created
Serverless Pool is only a computing engine and hence cannot store any data
Synapse SQL Architecture - Azure Synapse Analytics | Microsoft Docs