An Apache Spark-based analytics platform optimized for Azure.
Found the issue. My problem statement was incorrect. I thought I was querying the same connection in all 3 cases - notebook, query editor, and catalog browser. Instead I found that any connections I created that point to load-balancer are not working via serverless - this was due to NCC resource allowed was pointed at primary node rather than load-balancer. Where my connection is set to primary-node I can use serverless regardless of interface/method. This issue is due only to NCC configuration I set up was not matching connection definitions.