It's not SQL directly (it's the Kusto query language) which is more geared towards reading data and analytics.
I'm suggesting it however because the queries in it are also basically relational algebra (aka much of the knowledge is transferable to SQL - minus some special features like anomaly analysis), they give decently sized datasets with billions of rows and the challenges themselves were fun.
It's not SQL directly (it's the Kusto query language) which is more geared towards reading data and analytics.
I'm suggesting it however because the queries in it are also basically relational algebra (aka much of the knowledge is transferable to SQL - minus some special features like anomaly analysis), they give decently sized datasets with billions of rows and the challenges themselves were fun.