+ /* Yeah yeah, i know, floats are ew.
+ * We used them here because we'd be casting to float anyway to do this maths,
+ * and also only floating point numbers can deal with the pretty large numbers
+ * involved in the total throughput of a server over a large period of time.
+ * (we dont count 64 bit ints because not all systems have 64 bit ints, and floats
+ * can still hold more.
+ */