When monitoring Azure Stream Analytics job performance to ensure adequate streaming unit (SU) provisioning, Backlogged Input Events and Watermark Delay are the two most critical metrics to monitor alongside SU % Utilization.

Why Backlogged Input Events (A) is essential:

• This metric indicates the number of input events that are waiting to be processed by the job
• A consistently high backlog suggests the job cannot keep up with the incoming data rate
• Growing backlog indicates insufficient processing capacity, requiring additional SUs
• This directly correlates with whether the current SU allocation can handle the workload

Why Watermark Delay (B) is essential:

• Watermark delay measures how far behind real-time the job's processing is
• High watermark delay indicates the job is falling behind in processing events
• This metric shows temporal processing lag, which directly impacts job performance
• Sustained high delay suggests the need for more SUs to catch up with real-time processing

Why other options are less suitable:

• Function Events (C): While useful for monitoring Azure Functions integration, it doesn't directly indicate SU capacity needs
• Out of Order Events (D): This relates to event ordering and windowing logic, not capacity planning
• Late Input Events (E): This indicates events arriving outside their expected time windows, which is more about data quality than capacity requirements

Monitoring Backlogged Input Events and Watermark Delay together with SU % Utilization provides a comprehensive view of whether the job has sufficient resources to handle the workload efficiently.