Question 23
Domain 6: Evaluation and MonitoringWhen you enable AI Gateway inference tables on a serving endpoint, what table do you point Databricks to for logging?
Correct answer: B
Explanation
AI Gateway inference tables are enabled for logging, so you do not point Databricks to an existing table. The exam guide says to “Use inference tables and Agent Monitoring to track a live LLM endpoint,” which implies Databricks manages the logging table in Unity Catalog rather than requiring a precreated one.
Why each option is right or wrong
A. An existing Delta table that you create ahead of time
Inference tables are Databricks-managed in Unity Catalog, not precreated by the user.
B. No existing table; Databricks creates a new Unity Catalog inference table
The exam guide explicitly ties this objective to “Use inference tables and Agent Monitoring to track a live LLM endpoint,” which means the logging destination is managed by Databricks rather than manually selected by the user. Under AI Gateway inference tables, Databricks provisions the logging table in Unity Catalog automatically when you enable inference logging on the serving endpoint, so there is no preexisting table to point to.
C. A DBFS path that the endpoint writes JSON files into
AI Gateway inference logging uses inference tables, not DBFS JSON file output.
D. A system table selected from Catalog Explorer
Logging is configured through inference tables, not by selecting a system table in Catalog Explorer.