Question

I have a database in MongoDb that contains two collections: ‘categories’ and ‘articles’.
I’m using Mongoose on NodeJs to connect to the database and read the categories. I want to calculate the number of articles for a category without making an additional request/query, so it would be perfect if I could solve this at the database level.

An item from the ‘categories’ collection looks like:

{
'_id' : ObjectId("..."),
'feed_id' : 1,
'name': 'Blog posts'
}

An item from the ‘articles’ collection looks like:

{
'_id' : ObjectId("..."),
'feed_id' : 1,
'title': 'Article title',
'published' : '12/09/2012',
...
}

so the categories and articles are linked using the ‘feed_id’ field.

I would like to export all categories together with a corresponding number of articles:

{
'_id' : ObjectId("..."),
'feed_id' : 1,
'name': 'Blog posts',
'no_articles': 4
}

I’m not sure how exactly I should do this:

1) Create a ‘no_articles’ field in the categories collection? If yes, I would like this to be updated automatically when a document is inserted or deleted from the articles collection.

2) Sum up the articles into ‘no_articles’ when categories are read?

I read something about MapReduce and group, but didn’t quite understand if it’s possible to use them for this particular task.

Answer 1

This is one of use cases where traditional relational databases really shine.

It is impossible to do that with one query in mongodb. The “no_articles field” you mentioned is the way to go. Common name (among Rails people, anyway) for this approach is: Counter Cache Column. I am not very familiar with Mongoose, so I don’t know whether it will maintain that field for you or not. MongoDB itself certainly won’t do it. But maintaining it yourself isn’t a lot of work, you just need to be accurate.

I advise against counting articles when you read categories. This is a classic example of N+1 query problem and counter cache column is there to prevent it.