This page describes a technique of measuing the popularity of cache objects.

Zipf's Law has been applied to the Web with good success. It states that the request frequency for a Web object is inversely proportional to the object's popularity ranking. If you plot the logarithm of request freqency on the Y-axis and the logarithm of popularity ranking on the X-axis, the data fits a straight line. For example, see these plots. The slope of this line is the exponent in the formula:

Below we plot the exponent values for the NLANR caches over time.

The exponent values are calculated with this program: popularity-index.c