[tgv]

I've learned and applied my schema normalization and what have you got. But it's not the be-all and end-all of good engineering. What I greatly appreciate about hierarchical value storage in contrast to related flat records, is that it is so much easier to store and retrieve a tree. No need to generate ids and insert rows one by one, no need to decode the result of large joins. Because it doesn't only take time to write code for that, it can contain errors too.

If you've got hierarchical data and you just want to store, update and retrieve it as a whole (which is my use case), JSON is a good choice. Granted, it could be stored as a string/blob in my case. I don't really need to search within.