I just browsed a couple of old sweetcode.org articles and stumbled onto the "Judy functions" located at http://judy.sourceforge.net/.
Snippet from sweetcode.org's description: "Judy is an implementation of a sorted associative array in C. All reasonable associative array implementations have O(log N) insert and lookup complexity, but the constant factor for Judy is very much smaller than other widely used implementations (several times faster than STL's "map" and twice as fast as SGI's "hash_map", in my trivial benchmarks)."
This made me think of how beneficial it would be to Pike.
There is a lot of doc's at the site, man pages, pdf-documents with both brief and in-depth descriptions and analysis.
On associative arrays: http://judy.sourceforge.net/application/judysl.pdf
On scalable hashing: http://judy.sourceforge.net/application/Judy_hashing.pdf