undefined | Better HN

0 pointsjudofyr6y ago0 comments

> This is non-sensical.

Your examples are kinda proving yourself wrong though.

> Take a highly nested JSON document and try and implement it in a relational database. You would have an O(1) lookup in MongoDB and a O(n) lookup in MySQL.

In PostgreSQL (which is not MySQL, but it's a relational database) you can create an index for a column which contains JSON and the lookup for a nested field becomes O(1).

> Or good luck traversing a graph structure in MongoDB when you should've used Neo4J.

Facebook implemented the social graph on top of MySQL: https://www.facebook.com/notes/facebook-engineering/tao-the-....

> So in order to have a performant database you do need to "pay attention" and ensure that your access patterns suit the database type.

I think the point of the parent poster was that you don't need to ensure that your access pattern suite the database type, but that it fits well with the features the database provides. If you're looking at the database landscape today talking about a "database type" isn't really saying much at all.

0 comments

threeseed6y ago

That's because you took my examples and distorted them.

I specifically mentioned MongoDB and MySQL. Having a JSON type is basically a mini document database since it uses a different/enhanced query language and is often not relational with the rest of the schema.

I also specifically mentioned MongoDB and Neo4J. Yes you could implement a graph in MySQL but as your paper lists you will need to front it with a Memcache in order to cache the graph traversal lookups. With a graph database you generally don't need such a cache.

j / k navigate · click thread line to collapse