Hacker News Reader

Whilst I was an intern at Google I worked with Hilbert Curves. Range queries are something BigTable and Cassandra excel at and are the core database component of any Hilbert Curve based spatial index.

The real issue is building the database queries. Google released Uzaygezen[1], a Java library for multi-dimensional indexing, that handles the majority of this for you.

[1]: http://code.google.com/p/uzaygezen/

View on HN

I did not get this in the article - how exactly do we construct range queries (which the Hilbert curve improves on) here? Is the quadtree range resolved into actual coordinates, or is the datatype of the index bound to be integer, so we can query a range over the index (0110110 to 0110111 = 54 to 55, so the query range is [54, 55])?

Thanks Smerity that was really interesting - we are looking at applying special searches to Riak.

Reply on news.ycombinator.com