I have written a small application which loads artist information from Last.fm and saves JSON results in Couchbase. Application starts with on arbitrary artist, loads it’s information and continues recursively with artists similar to the this one. This process never stops, since every artist on Last.fm has similar artists. To speed up performance loading and saving is done in multiple threads. After running the application for one hour you have thousands of artists in your Couchbase bucket. Running it for a longer time it will produce a kind of “Big Data”, useful data, not clutter created by a simple loop.