https://docs.microsoft.com/en-us/azure/databricks/integrations/bi/microstrategy

Your proposed solution to have 2 containers (1 for real time and one for long term) is certainly possible, but not required.

You could land ALL data into a bronze layer (streaming or not) for example. And move on from that layer.

The main attention point is that you should try to minimize the query load on the source system (so avoid ingesting the same data twice).