- Identifying the data elements and sources for that data
- Coming up with logical data model that will support the business’s analytical needs
- Often, getting this logical model; or as technical teams would say LDM (Logical data model) approved through some governance body in the organization
- Getting the physical data structures, PDM (Physical data model) created
- Extracting and transforming data from various sources to load into the physical data structures
To summarize, a typical data warehouse build out contains following workflow (of events):
Typical workflow for a Data Warehouse build out [Please note that we will be modifying this as we discuss the topic further] |
Now, let’s talk about Agile data warehouse (ADW). By ADW, I mean:
- We build the data warehouse in increments
- Delivering Potential shippable increment (PSI) of the data warehouse at a regular frequency [and not have customers waiting to get the data warehouse as one big-bang delivery]
- Involving customers through the build process
- Understanding customers, and their needs – what are the business reasons for them to request this data warehouse
- Focus on creating customer value incrementally, and not the technologies or the implementation of it. Shift your focus from technology (and data elements, data structures, primary key, foreign key, etc.) to customers needs.
Often times data folks are so focused on their tables and primary keys that they start driving the build out from that vantage point only 🙁