Ask HN: When do you use SQL vs NoSQL?
Some people like SQL, some like NoSQL.. some have used PostgreSQL and moved to MongoDB, and some vice versa. But at the end of the day, it usually comes down to "use whatever you're comfortable with"..
So I'm wondering, what data patterns are more appropriate to SQL vs NoSQL datastores? Can such a decision even be simplified to 'data pattern'?
1 comment
[ 3.7 ms ] story [ 13.9 ms ] threadI generally use SQL when I have really well defined relationships that change rarely and where I have data that needs to be shared amongst many disparate systems (especially if I don't control one or more of them).
NoSQL I use more as an application specific data source and cache. Many times this means warming up the NoSQL from the relational side for some data elements. I really don't like to pollute a SQL datastore with application specific components unless they need to be shared and I don't control all the applications.
I don't agree with people that draw the line at the need for transactions. Transactions can be managed in either SQL or NoSQL and even across them both with some care.