You are monitoring an Azure Stream Analytics job.
You discover that the Backlogged Input Events metric is increasing slowly and is consistently non-zero.
You need to ensure that the job can handle all the events.
What should you do?
Click on the arrows to vote for the correct answer
The "Backlogged Input Events" metric in Azure Stream Analytics job indicates the number of input events that have not been processed due to either the job being busy or encountering issues in processing. If this metric is increasing slowly and consistently non-zero, it indicates that the job is not able to keep up with the incoming events and is falling behind.
To ensure that the job can handle all the events, you can take the following steps:
Scale up the job: You can increase the number of Streaming Units (SU) assigned to the job. Each SU provides additional resources such as CPU, memory, and I/O bandwidth to the job, which can help it process events faster. To scale up the job, go to the Azure portal, select the Stream Analytics job, and increase the number of SUs assigned to it.
Optimize the query: The job's query may be inefficient or processing events in a suboptimal manner, causing it to fall behind. You can optimize the query to improve its performance. Consider reviewing the query logic and ensuring that it is using appropriate windowing and aggregation functions. Also, check if the query is optimized for the input data schema.
Use a partitioned input source: If the input source is partitioned, you can enable partitioning in the Stream Analytics job to improve its performance. Partitioning allows the job to process multiple partitions in parallel, thus increasing its throughput.
Increase the number of input partitions: If