Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> But the very title of Codd's paper mentions shared data banks, doesn't it?

Shared as in on a single mainframe, yes. Remember, Codd was a mainframe guy — even helping to design one before starting work on the relational model. It wasn't until Oracle came along did anyone really think it would be a good idea to slap networking directly on top of SQL.

> Concurrent access and thus networking was there from the beginning.

Again, the trouble is the high latency that networks introduce. That wasn't there from the beginning. Mainframes are designed for low-latency. That is a relatively new constraint that Codd didn't need to think about...

...But we in the internet age do. The high latency means that, in many cases, we can't realistically use SQL as it was intended. Which is why we have all ended up building bespoke DMBSes that speak things like JSON instead of SQL. Granted, there have been some efforts to bring richer data structures to SQL, including JSON support, but they're all pretty hacky, frankly. More ideal would have been to design a better language for the client/server database model from the start, but we are no doubt in too deep now. Worse is better applies.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: