> "Suppose you want to write a database. You'd probably start by implementing relational algebra operators — projection, filter, join, etc. The easy way is to implement them as functions that take in tables and return tables, and assemble them into a larger expression. That was how Prela worked in its first incarnation. The code was clean, but it was hella slow! Which was not surprising, because every operator materialized every intermediate result. "

DB development starts with the storage engine, file manager, buffer pool (page cache), and page access methods (heaps/indices) which are binary buffer views. Then, you add the transaction manager, the WAL/recovery bits.

The actual implementation of relational algebra and a SQL language + parsing are little icing layers on top of a transactional storage engine.

https://news.ycombinator.com/item?id=48356563