You are configuring Azure Synapse Analytics pools to support an Azure Active Directory group (Group3) with the requirement to 'Query and load data from Apache Parquet files stored in Azure Data Lake…

Apache Spark pools are for big data processing and ML, and while they can query Parquet files, their cost model is based on compute hours, not strictly on data processed, and they require provisioning.

Dedicated SQL pools are provisioned data warehouses with fixed compute costs (DWUs) and are not designed for direct querying of files in a data lake with a pay-per-query model based on data processed.

Serverless SQL pools are purpose-built for ad-hoc querying of data directly in data lakes, supporting formats such as Parquet. Their cost model is consumption-based, meaning you only pay for the data processed by your queries, which directly aligns with the requirement for costs based on the amount of data processed.