We’ve been using using Looker successfully for the past 6-8 months, but are now outgrowing the initial design, and want to create a more powerful Data Warehouse – more data in it, more analyses supported, etc. As part of the process, my team and I are reading the classic(?) Kimball’s Data Warehouse book.
My question is pretty general at this point: Do you find that you need to follow all the techniques in that book strictly? Or do you find that because Looker has things like symmetric aggregates and lets you explore complex joins more easily, you are breaking some of the “rules” in the book? For example, I’m wondering about strictly sticking to a star schema vs. doing some snowflaking? i.e. Does that book need an update because of what Looker has done?
If we follow the Kimball approach, at least on the face of it, it looks we might have to do a lot more in ETL than I did in the first iteration of the data warehouse. Maybe that’s what Looker does? Allows for simpler ETL, since more can be done inside the LookML layer? Do you have any examples of this you can share?
I will likely have more specifics to ask, but wanted to open with this one!
Thanks so much,