Why Azure Functions Event Hub Trigger not utilizing all host instances to process the messages?

Question

We have the following: - Azure Functions Event Hub Trigger Function - Language/Runtime: Python 3.9 - Resources are in the same subnet of the Virtual Network - Allowed access based on User Assigned Managed Identity - Azure Functions Premium Hosting Plan with event hub 32 partitions We identified that there are some unused host instances where few hosts used 100%. ![221852-uqtyz.png][1] Do anyone light me that why the Azure Functions Project is not utilizing all the hosts to process the messages/load from the Event Hub? [1]: /api/attachments/221852-uqtyz.png?platform=QnA

Answer 1

anonymous user ,
Premium plan functions will make use of the other instances if the scale controller determines the instance(s) currently handling the events aren’t able to handle the load, regardless of the pre-warmed and minimum instances. So even though you have 5 in pre-warmed and minimum, it’ll start sending all the events to one instances and then load balance to the others once that one starts to be overwhelmed. This is how the scale controller load balances. If you look at the cpu utilization, the average cpu utilization remains at 67% and 57 %, only when they exceed 70% and remain for some time, the events will be routed to other instances.