Nice to see so many Thomas Neumann paper references :) Umbra (paper: pdf) is Thomas’s current DB system where an evolution of LeanStore is applied to get in-memory DB performance while spilling to disk properly.
What does this actually do? The reasons explains how to compile, how to launch a shell, etc. But it does say what this thing is other than “SQLite for analytics”. What does that mean specifically?
I’m looking for differences between duckdb and sqlite and I’m not seeing any writeups to clearly show why one is better for certain use cases over the other
Also, after reading the docs:
This article discusses OLTP vs OLAP with a focus on DuckDB and SQLite.