Bottom-Up Data Warehousing
Have you been tasked with building a new data warehouse for you company? Is the existence of the data warehouse your only requirement? Are you having nightmares about the enormity of this task and the resources that will be needed? You may want to consider a bottom-up approach to building your data warehouse.
From Data Mart to Data Warehouse
Rather than spending months gathering all of your requirements and then proceeding to build a massive solution (using standard waterfall development) a bottom up approach will begin with basic data marts designed for functional areas of your business.
Develop Functional Data Marts
Data marts can be designed and developed in a relatively short amount of time. This means that your business users can start benefiting from reporting solutions in a matter of weeks rather than months. You will receive multiple benefits that are expected when using an agile development style.
Document Exposed Requirements
One of these benefits includes gathering data warehouse requirements through exploration. Users will be able to explore your reporting solution and they will be excited to browse the data using self-serve data tools. Be sure to pay close attention to the questions that your users begin asking. These questions are sure to become requirements for your data warehouse solution.
Once users are comfortable I recommend that you encourage forward thinking by helping to prompt “what if” questions. As an example, “What if you could view sales trends on the same chart as product enhancement releases? Would that data be useful?” You may even choose to accomplish this in official requirements meetings or just have short conversations as you deliver new reports.
Create Conformed Facts and Dimensions
This step will not happen until well after users have successfully adopted your reporting technologies. Quality documentation is key to accomplish this task. Each existing data mart will need to be cross referenced to identify commonalities between your measures and dimensions in your various data marts.
Conformed dimensions and facts will be used to create consistency for the foundation of your data warehouse. You will ensure that measures have identical definitions in every data mart that they exist in.
Data Warehouse Development
Armed with your detailed requirements documents and data mart documentation you are now ready to establish an official data warehouse. You have already done all of the heavy lifting with your data mart ETLs. Now you will just build the structure that merges data from multiple functional areas in your business. The provides the much sought after “One version of the truth” that so many companies are seeking.
Thanks for reading!