Top 10 things you must know before designing a data warehouse
ETL solution takes more time to design than analytical solutions
Needless to say, these assumptions often turn void when it comes to actually coding the ETL layer to feed the data warehouse. Almost always, mapping, cleaning and preparing data turns out significantly more time consuming compared to design of Reporting/Analytics layer.
2. Data Warehouse scope will increase along the development
Users often do not know what they want from the data until they start to see the data. As and when development progress and more and more data visualization becomes possible, users start wishing even more out of their data.
.. Many requirements that were implicit in the beginning becomes explicit and indispensable in the later phase of the project.
3. Issues will be discovered in the source system that went undetected till date
.. 6. Users will create conflicting business rules
.. And when you take the data from all these departments and try to combine them together into an integrated data warehouse, you would often discover that business rule suggested by one user is completely opposite to the business rule suggested by the other.