Storing arrays in the database

Question

I'm wondering if it is actually good practise to store Arrays in the database ? I tend to use json_encode rather than serialize, but was just wondering if it is a good idea. If not, then I can make some small changes and just implode the array with a comma.

Answer 1

No, it's a terrible practice. Please refrain from inserting CSV, JSON*, serialize() or ANY kind of serialized data in a relational database. Denormalization is almost always a bad idea - don't do it unless you really know what you are doing, or you'll start asking questions like: this, this, this, this, ...

Doing that, you lose or it severely hinders your ability to:

• Use JOINs.
• Find or modify a particular element
• Enforce referential integrity
• Benefit from index usage
• And it also wastes space

It may sound pedantic, but seeing people do this is one of my pet peeves - especially in light of the plethora of questions asked on SO that would be avoided if they did the right way.

Here's the right way to do one-to-many and many-to-many relationships in an RDBMS.

^{*Although some SQL databases have built-in support for JSON, it's often better to restructure your data so that you don't need this}

Answer 2

Depends on your usage pattern. If you're going to need to access smaller portions of the array (e.g. for use in a where clause or similar), then it's a bad idea - you lose all the benefits of storing data in a relational database by making the data un-relatable. You'll end up with major overhead extracting that small piece of data over and over and over again.

On the other hand, if you're just using the database as a data store and never need to slice that stored array apart - just insert and retrieve, then there's probably no problem at all, other than maybe waste of space, as a serialized/json'd format tends to be "wordy" and take up more space than the raw data itself does.