What not to do – anti-patterns for a data warehouse

While there are many good ways to use a data warehouse for analytics, there are some things that organizations may be tempted to do that are not good for a data warehouse.

Let's take a look at some of the ways of using a data warehouse that should be avoided.

Using a data warehouse as a transactional datastore

Data warehouses are designed to be optimized for online analytical processing (OLAP) queries, so they should not be used for online transaction processing (OLTP) queries and use cases.

While there are mechanisms to update or delete data from a data warehouse, a data warehouse is primarily designed for append-only queries. There are also other features of transactional databases (such as MySQL or PostgreSQL) that are available in Redshift – such as the concept of primary and foreign keys – but these are used for performance optimization and query planning and are not enforced by Redshift...