CouchDB stores schema-less JSON documents in a flat namespace. Each document has an id that's either auto generated or provided by the user. The JSON documents can be as simple or as complex as it can be. You can think of each document as an instance of a particular class. CouchDB doesn't know or track relationships between documents. It's up to you to keep track of that.

The sample DB below contains information about a bunch of pictures uploaded and tagged by different users with some meta data about each picture.

Now that we have the documents in place, we can use the map function to generate complex key value pairs (sorted by key) and then reduce the values corresponding to each unique key into either a simple or complex value.

This is a very key concept so it's worth reiterating. The key/value pairs generated by the map function are stored in a BTree and are updated incrementally as documents are added or updated. The query against a view is simply slicing this sorted key space, potentially reducing the slice and returning the values. CouchDB will cache the reduced values in the BTree as well and will try and reuse these cached values during the key-space slicing, a.k.a. query.

Try and learn: choose one