What are the different approaches for auto scaling (out & in) for Iot Hub

Question

Hello,

I have a requirement for dynamic scaling of IoT hub based on condition like whenever quota exceeded, scale up and when all normal, scale down.

Sometime some of the devices sending chatty messages, I also required a way to find out that device and accordingly do some business logic around it.

I went through below durable azure function approach which is quite old, https://github.com/Azure-Samples/iot-hub-dotnet-autoscale

Looking for some suggestion and approach here? Thanks and Appreciate!

Answer 1

Hello @@IamCoder ,

at this moment, the IoT Hub only supports manual scaling but this mechanism can be automated.

The code sample is a solution based on this mechanism and is explained here.

This is still the preferred way to do this although you can build a similar solution independently.

An IoT Hub ingests a number of messages (or better: 4Kb chunks) per day and based on that, the scale needs to be set. The number of devices is not a factor.

See here for the amounts of messages per tier.

Next to the regular amount of expected messages, it's also smart to check for expected connection loss.

This means, if devices are able to store messages locally in case of connection loss, these will be send afterwards.

So, if the connection is lost for, say 24 hours, you can expect the double amount of messages the next day (in this example with a TTL set to 1 day on the device).