Because of the dynamic nature of the software system, with
complex interrelationships between page management, real storage
allocation, and program behaviour we decided to use sampling techniques
to determine the status of the page queues and the frequency of page
faults.