My organisation is implementing a new data warehouse using a columnar data store (Redshift) and the performance is currently abysmal. There are many causes for this but I think the key reason is because of the way we have our data structured, we're implementing a classic star schema design, which surely is optimal for a traditional RDBMS (OLTP) and not necessarily for OLAP?

I'm looking for someone with experience on this to briefly explain the pros and cons of structuring large quantities of data in a star schema on amazon redshift.

I'm obviously no expert, but I'm looking for some advice and guidance on this.

thanks, Matt

2 Answers 2


You are mixing up three orthogonal concepts: data model (star schema), workload characteristics (OLTP vs. OLAP), and physical data organisation (columnar).

Your data model has no bearing on whether column-organised tables are appropriate for you; however, data organisation must reflect the nature of your workload (i.e. queries). For example, SELECT * FROM... cannot possibly perform well in a columnar database.


Mustaccio's answer is good. In addition, every star schema I have ever seen was implemented on top of a relational DBMS, often Oracle. Also, the data in it was downstream from some kind or operational database, also on a relational DBMS. The operational database was mostly normalized and used for OLTP.

I have never seen a star schema used for OLTP. I can't imagine anybody ever doing that. It's not what star schema is for.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.