The Elusive Idempotent Data Load/ETL

This is a topic I’ve been musing about lately. The idempotent data load has been a source of much pain and suffering in the lives of many a data engineer and data warehouse developers. Apparently somethings don’t change with the passage of time. My first job in tech was working on a data warehouse team with a classic Kimball style model on SQL Server, back then worrying how to make data loads and ETL idempotent was the task of the hour. All these years later working on data lakes in DataBricks with Spark … guess what …. still worrying about idempotent ETL and data loads.

Read more

Data Modeling in DeltaLake (DataBricks)

Time to open a can of worms. I’ve recently been working with DataBricks, specifically DeltaLake (which I wrote about here). DeltaLake is an amazing tool that when paired with Apache Spark, is like the juggernaut of Big Data. The old is new, the new is old. The rise of DataBricks and DeltaLake is proof of the age old need for classic Data Warehousing/Data Lakes is as strong as ever. While this Spark+DeltaLakes tech stack is amazing, it’s not your Grandma’s data warehouse, it’s fundamentally different under the hood. One of the topics I’ve been thinking about lately has been data modeling in DeltaLake (on DataBricks or not).

Read more