4 ms·
How are these issues any less for NoSQL databases? They don't have non-obvious idiosyncracies? You don't need to understand their concurrency models?
by noobar 6y ago
How are these issues any less for NoSQL databases? They don't have non-obvious idiosyncracies? You don't need to understand their concurrency models?
- dathinab 6y agoThe problem is if you don't know about a lot of SQL complexity and you are not aware of you not knowing it really easy to end up with a major problem with "corrupted" (here incorrect) data while the main reason which is often cited for RDB is to prevent exactly that. This can easily lead to an situation where people (wrongly) come to the conclusion that RDBs don't work anyway and they should just go to NoSQL DBs as it's "just easier" to not have to handle a scheme. Which yes is completely flawed as a argument as doing NoSQL without any implicitly schema can bit you very hard and you still want to do transactions but now your DBMS might not support the transactions you need (instead of you just doing them incorrectly) so you "somehow" bolt them on top of it and if it fails now it's "clearly your fault". Except there is a good chance it isn't as some NoSQL DBMS are much more tricky to handle then they seem (like Mongo not necessary keeping to the Durability part in ACID |EDIT: In the past I don't know about now). I have seen this fallacy or parts of it more then one time both in "real live" and comments on Hacker News.