Question

You manage a Direct Lake semantic model on a Fabric F64 capacity. Business stakeholders complain that report performance is inconsistent: some visuals return instantly while others are noticeably slower, especially during peak hours. Investigation shows that when the model exceeds the maximum in-memory row guardrail, queries silently fall back to DirectQuery against the SQL analytics endpoint, causing the slowdowns. Leadership decides they would rather have a query fail with a clear error than silently degrade performance, so they can identify and fix the offending report. Which configuration change on the semantic model achieves this?

Accepted Answer

Set the DirectLakeBehavior property to 'DirectLakeOnly' so queries never fall back to DirectQuery and instead return an error when guardrails are exceeded

Answer

Set the DirectLakeBehavior property to 'DirectQueryOnly' to force all queries through the SQL analytics endpoint

Answer

Enable the Large semantic model storage format to raise the in-memory guardrails and prevent any fallback

Answer

Configure incremental refresh on the fact table to reduce the number of rows loaded into memory