There is a group of sensors that send data to Azure Event Hub in real-time. This data is then sent to the Azure Stream Analytics job for processing. You are setting up a metric-based alert system whenever there are more volume inputs, and stream analytics resources cannot handle it. Which of the following metrics should be used to set up this alert?

Answer options:

A.Watermark Delay
B.Out-of-Order Events
C.Early Input Events
D.Data Conversion Errors

Correct Answer: A
Watermark delay metric is used as a source to find out the job health in Stream Analytics. It is calculated by subtracting the largest watermark that happened till then from the wall clock time of the processing node. A small illustration of this is given in the picture below. 

Image source: Microsoft Documentation 
When there is a resource scarcity in the Stream Analytics job, the watermark delay value also increases. So, if you set up an alert based on the threshold in this metric, the notifications will be sent.
Option A is correct: Based on the watermark delay metric, the job health can be monitored. In the case of resource scarcity, the watermark delay metric increases.
Option B is incorrect: This metric shows the number of events that arrived out-of-order and then dropped/ adjusted the timestamp.
Option C is incorrect: This metric shows the events that arrived earlier and then dropped or timestamp got adjusted.
Option D is incorrect: This metric shows the value of events that were not converted to the desired schema after processing.
Reference: To know more, please refer to the docs below: 

https://docs.microsoft.com/en-us/azure/stream-analytics/stream-analytics-time-handling#watermark-delay-details
https://docs.microsoft.com/en-us/azure/stream-analytics/stream-analytics-monitoring