I hope you are asking for a SQL version of the pyspark code ? Can you please explain the advantages of having it in SQL as compared to pyspark ? There are some options, best would be federated queries against sql server ? select as if it were a databricks table and write it to the target ? alternatively you could create a view against the sql server table and then use that as insert into the databricks table...but in my limited understanding (could be wrong)  all of it would basically get optimized similarly in the background and serves no additional benefits to be rewritten in SQL...