MicrosoftThanks for a terrific article. I do have a few observations and comments. It seems like the assumption is a simple view is all that is needed in the silver layer which I don't agree with. Wouldn't the silver layer contain canonical delta tables that have core transformations, quality checks and de-duplication of data which is key to merge, and canonical data sets? I would also think that a bronze shortcut would be preferred over a view so they are available to us in a Data Factory Copy Data activity. By having canonical delta tables and bronze shortcuts in our Silver Lakehouse we maintain the direct lake capability opening up options for quality checks and PBI connectivity. When building the Gold Lakehouse why wouldn't you just create delta tables that represent your star schema data warehouse? Since the Lakehouse has a SQL Endpoint all sql users/toolsets could access it that way and anyone needing to run AI/ML on the company vetted data warehouse could do so using the delta tables/python/etc. I would really appreciate any/all feedback on these design thoughts. What's good or bad? What else should I consider? Thanks Everyone!
