If you have keys that tend to have common prefices (e.g. URLs in RDF) then checking them in reverse (e.g. with a suffix tree) is generally faster, as you will hit differences quicker.