Best Practice For Swapping Dataset Sources?

I have a few QuickSight Datasets all sourced from the same DEV SQL database. Since QuickSight is touted as an Enterprise BI tool, I am trying to find best practice documentation or any pointers on how to efficiently deploy my current Analysis/Dashboard/Datasets bundle, all sourced from a new PRD SQL database with all same tables/view entities.

Is duplicating the datasets, and re-pointing them to PRD database source the easiest thing to do?

The easiest thing to do would be to create the PRD data source and the PRD Datasets you are going to use and swap them at the analysis level.

Once you have the PRD datasets, you can go to your analysis and replace the current dataset with the PRD dataset you just created with the replace option.

Whether you want to create a new analysis or not is up to you. You can make copies of them easily.


Finally you would publish the analysis with the new dataset as a dashboard and that would complete the swapping.

Thank you. That makes sense.

1 step further… So that also means I could copy my Analysis and have 1 Analysis pointed to DEV, and the other Analysis pointed to PRD… but then If I need to make a development change, I will either:

  1. have to make it to both Analysis versions
    OR 2. make the change to DEV Analysis first, delete current version of Analysis PRD, then copy DEV Analysis, rename to PRD and swap sourcing back to PRD dataset(s)?

So there is no global parameterized way to elegantly swap database sourcing for multiple datasets from the same database source at the same time? It should be handled at the QuickSight dataset level is what I am understanding?

Yes that is correct. You would have to make changes to both or change the DEV and then make a copy from that.

However, you can think of the analysis like a dev env. That is where you test out how visuals / sheets work and look. The dashboard you can image it being like PRD because it will only update when you say publish as dashboard.