A simple DBScan Clustering Implementation
Note as of 8/31/2016 this is in heavy cleanup and rejiggering from some very old, messy code. If by some chance you've stumbled on this now, bear with me, it should be getting improved and cleaned up in the nearish future. That said, it seems to actually work. Ish.
Please note, there are also plenty of very nice (i.e. more effective/efficient) DBScan implementations out here. This guy is more of an exercise and something to be relatively clear and easy to read. If you use it, awesome!
A more crude version of it that was the basis of this was actually used in the published research, see the MASCOTS 2013 paper "Validating Storage System Instrumentation"