[redcalx]

For one, a database may receive a query such as "WHERE x >=10 AND x <=100". I.e. having an ordered index is useful for accessing ranges, whereas a hash lookup is always just for a single entry. The question then becomes - why are lookups in RAM more likely to be single lookups rather than ranges (and stats on ranges)? Partly at least because DBs provide a query language that make range based queries easy to do, and just because that's where most business data sits (most of it necessarily needs to be in a persistent store).

[jamesfmilne]

Everything is driven by the business case. Is there a business case to optimise for range queries? If so, optimise for that. Is there a business case for individual record queries? Then optimise for that. Is there a case for both? Then optimise for that.

The biggest lie of the 21st century is convincing JavaScript/Ruby/Python/Clojure/whatever programmers that web development is something sexier/holier/worthier than boring old CRUD Oracle Forms database development.

Its exactly the same but with fonts and animated transitions.

[zadokshi]

Oracle forms vs modern web development is a little Off topic but I’ll bite.

Could gmail have been built in Oracle forms, how about slack? How about Asana? How about an LMS such as blackboard, moodle, or D2L? Can you name any popular mainstream product that could be built upon and run off the Oracle forms product?

Oracle forms will get you 80% of the way there, and the final 20% will be impossible.

[garmaine]

> Could gmail have been built in Oracle forms, how about slack? How about Asana? How about an LMS such as blackboard, moodle, or D2L? Can you name any popular mainstream product that could be built upon and run off the Oracle forms product?

Yes, it could.

Not going to judge whether it's a good idea or not, but it most certainly could be implemented.

I suspect you're not aware just how much capability relational database management systems have.

[thu2111]

Don't know why this post was voted down. I used to work on Gmail, which uses a totally proprietary storage and database stack, like everything else at Google.

One of the most insightful design docs I ever read was an exploration by an engineer of how GMail 1.0 could have been implemented with commodity tech and how it'd compare cost wise. The rather sobering conclusion was that it'd have similar functionality and been cheaper to develop/run.