SDLC

I have spent some time doing this for customers. Thus far it has been a bit frustrating. I have had some limited success in simply editing the underlying data source to change it from dev to prod for example, but have run into problems as well. Generally, the problem is with adding calculated columns. Simply stated, if you hope for this to work, minimize or eliminate your use of custom columns and possibly measures, they often seem to cause problems when you swap data sources under the hood. You end up putting a lot more of this in M if at all possible, which makes your M code a lot more complex.

The really frustrating part with this is that you have to blow away the underlying data table and recreate it, which wipes it out of all of your visualizations, breaks relationships, it's really kind of a real mess.

One comment that I have on this because of the frustrations that I have had, if you can use prod as your starting point for your data model, by all means do that. I realize that this is not always possible because you are often changing the underlying data in dev or need stored procedures written, etc. but if that is not the case, use prod data, it's not like the load of data from prod into your model is a big performance hit on prod and you could limit the amount ingested at first and then remove that limit when your model goes to production.