We deployed Db2 Warehouse on the IBM Cloud backbone. We had a problem with our client because at first, they didn't want any warehousing in their solution. They wanted to generate reports and data analysis and predictions and forecast analysis for their company, but without any warehousing. They wanted to connect their ordinary relational database with IBM Cognos directly. The problem with that is that IBM Cognos Analytics doesn't work well with relational databases, so we suggested using some sort of warehousing database to make the Cognos engine work efficiently. They agreed and we started working with Db2 Warehouse.
The primary use case of this solution is reporting. Reporting doesn't work well with a relational database. Warehousing works with something called Star Schema or Snowflake Schema. Its structure is very different from the structure of a relational database. It consists of dimension tables and fact tables. All of your data is in your fact tables. It consists of one or two tables at most, but relational databases are not like that. Analytic engines don't work well with any relational data. Your report can be generated as fast as possible right now, but in three to five years, the report will perform poorly and there could be performance issues, so warehousing with reporting services instead is very good.
The analytics engine is not bad at forecasting predictions.