In previous post, we have discussed some ways to optimize Azure Stream Analytics outputs. In today’s post, lets have a look at streaming units which are used to optimize the performance of the Azure Stream Analytics jobs.
To optimize performance, we need to monitor and adjust the resources allocated to Stream Analytics jobs. Streaming Units are a way to measure this performance. This means, higher the streaming units allocated to a job, the more CPU and memory resources available to it.
SU% utilization metric is a measure of resource usage for Streaming units. This can range from 0% to 100%. It is recommended to keep the SU% utilization below 80% to accommodate future spikes and It is best practice to set an alert at 80%, so that more Streaming Units can be allocated to make sure that the job does not run out of resources. One indication that the Stream Analytics job requires more Streaming Units is a high number of backlogged input events.
Monitor Job Performance
Azure portal provides some metrics to measure the throughput of the job. One of the steps to improve the throughput of the job, that can be taken besides adjusting the partitions and optimizing the query is to add more Streaming Units to the job.
Reference: https://docs.microsoft.com/en-us/azure/stream-analytics/stream-analytics-streaming-unit-consumption
Azure Data Engineering 
2 thoughts on “Performance monitoring in Azure Stream Analytics”