@szymon_dybczak 
Thanks for your explanation. 

While I understand the limitations of Databricks as a OLTP read system, is there any solution at all that is read-optimized? Like an OLAP layer which optimizes for both aggregation and reads with low latency.