DuckDB is columnar, so in theory a lot faster than Postgres or SQL server for Analytical workloads.
DuckDB is to Clickhouse, TimescaleDB, Redshift, etc as SQLite is to Postgres, MySQL, SQL Server.
DuckDB is to Clickhouse, TimescaleDB, Redshift, etc as SQLite is to Postgres, MySQL, SQL Server.
Columnar technology in SQL Server only became usable in SQL 2016 (it existed in 2012 but was too restrictive — I know because I tried to use it).
In 2016 and above, you can either create a columnar index (non clustered column store) or convert an entire table into a columnar table (clustered column store). The technology is actually pretty impressive and I’ve used it in production where I have row stores coexisting with column stores within the same database.