The Backlogged Input Events metric in Azure Stream Analytics indicates that the job is receiving more events than it can process, leading to a growing queue of unprocessed events. This backlog occurs when the job's processing capacity is insufficient for the incoming event rate.

Why Option B is correct:

• Increasing the number of Streaming Units (SUs) directly scales the processing capacity of the Stream Analytics job
• Each SU provides approximately 1 MB/second of processing throughput
• Scaling up SUs allows the job to process events faster, reducing and eventually eliminating the backlog
• This is the primary scaling mechanism for handling increased event volumes in Stream Analytics

Why other options are not suitable:

• Option A (Change compatibility level): Compatibility level affects query language features and behavior, not processing capacity or throughput
• Option C (Use $default consumer group): Consumer group configuration affects how events are read from Event Hubs, but doesn't address insufficient processing capacity
• Option D (Create additional output stream): Adding outputs doesn't increase processing capacity for input events; it only creates additional destinations for processed data

The most effective solution for a consistently growing backlog is to increase the job's processing capacity by scaling up the Streaming Units.