We are developing an application where IoT devices will be publishing events to azure IoT hub using MQTT protocol (by using one topic to push message). We want to consume these message using Stream Analytic service. And to scale Stream analytic services, it is recommended to use partitionBy clause.
Since, we are not using Azure Event hub SDK, can we somehow attached partitionId with events?
Thanks In Advance
As Rita mentioned in the comments, Event Hub will automatically associate each device to a particular partition.
Then, when you can use PARTITION BY PartitionId for steps closer to the input to efficiently parallelize processing of the input and reduce/aggregate the data. Then, you can have another non-partitioned step to output to SQL sending some aggregate data.
Doing that you will be able to assign more thank 6 SUs, even with an output to SQL.
We will update our documentation to give more info about scaling ASA jobs and describe the different possible scenarios.
Thanks,
JS - Azure Stream Analytics