Comment by nickpeterson

A lot of oltp databases have modeling conventions for making read only reporting tables. Do tabular dbs have an inverse for transaction heavy data, that later gets batched to a read optimized structure? I kind of think most databases (even oltp workloads) really are read dominated. I feel like DuckDB is really close to working as the ‘main db’ for such systems, but my lack of knowledge for how to handle quick mutations bothers me. It feels like some form of temporal data modeling would solve it but I don’t know.