@Brett Gerhardi​ Did you get a satisfactory solution for this? We are also looking to migrate our warehousing to Databricks, but SCD 2 without effective surrogate keys is.... less than desirable.

@Tom Renish​ Thank you for offering this. Have you found this works reliably? It just seems strange that such a work around is necessary. Surely surrogate keys for SCD 2 are essential? Or are Databricks expecting us to use inefficient, convoluted joins with combinations of Natural Keys and dates?