It's very easy to underestimate how difficult building a database from scratch (or even making large incremental changes to an existing one) is. Another way to

It's very easy to underestimate how difficult building a database from scratch (or even making large incremental changes to an existing one) is. Another way to think of it is to think of an core database engine as analogous to a part of a propulsion mechanism. You can "demo" the core to show that a technology has promise. In case of a DB, you'd use a micro-benchmark with a synthetic workload; in case of a propulsion mechanism, you could use a software simulation. The full RDBMS (transactions and locking, query optimization, buffer pool, etc...) takes far longer to build. Finally, the equivalent of a "spaceship" here would not be more than the RDBMS, but a full ecosystem of systems software, middleware, and applications (e.g., app servers, messaging, CRM, ERP, etc..)