What is the best approach for adding reflections?


I would like to know the best approach in adding reflections?

Should we start at the lowest level? i.e at the PDS level?


It really depends on your situation… if you are using slow storage like JSON or CSV, then creating a partitioned reflection backed by parquet files will significantly improve performance.

Otherwise, if you trying to accelerate dashboards or reports, then look for common views at the top of the query tree to enable reflections on.

I have a very large table in azure Managed instance on which i have a view based on.
But when adding the reflection on the view, it is taking a lot of time to create the reflection.
Then this view is combined with other views in a single view to be used in the PBI report.
what would be the best approach in that case?

Sounds like you want to make the PBI report fast… in that case, I’d consider enabling a raw reflection on that single view.

If you were on Dremio Enterprise version, there are tools to make reflection recommendations across PBI workloads so that the queries could better share and re-use the reflections: