DWH could be thought of as a layer of abstraction between entities in IT systems and business entities. To make the process of filling up Data Warehouse painless there should be guidelines for how we store data in IT backends. Our experience suggests that correctly designed requirements may significantly increase the speed of DWH creation.

Classic DWH systems are designed to process data on the daily basis. This implies that end-users see the results on the next morning. From our practice we see more and more companies trying to observe same day situation. Some are comfortable with 1 hour of delay while others want to have real-time monitoring. These aspects should be thought of in advance.

DWH is a complex system in itself. Usually, there are at least 3 layers that exist. Raw layer is responsible for data ingestion from source systems be it production databases, event streams or spreadsheets. Detailed layer as its name suggests is made for convenient layout of data for Data Engineers. Marts is the layer where business entities are created for end-users.

Many companies that we worked with have very sophisticated system of managing access to data. 2 of the most popular cases of sensitive data include personal data of users and financial data of the company. Correctly designed DWH should have simple yet efficient approaches to implement various access policies.