While Ehcache concerns itself with Java objects, caching is used throughout computing, from CPU caches to the Internet Domain Name System (DNS) system. Why? Because many computer systems exhibit "locality of reference."

Data that has been recently accessed is more likely to be accessed soon.

The 80:20 rule, also known as Pareto distribution is a power law probability distribution used to describe a multitude of phenomena from various areas such as social science, sizes of objects in nature, stock prices, computer science, insurance, etc.

In simplified terms, and leaned towards the way that Ehcache exploits this distribution, this means that a small amount of data makes up for the most traffic and vice versa.

Digging deeper, this allows calculating from these 20% of data making up for most of the traffic the 20% responsible for the higher end of the traffic, and so on.

If you can only afford to cache a subset of your data, then an intelligent cache, such as Ehcache, can really help by keeping the hottest data available. However keeping all of it available is even better, and Ehcache has the features to be able to do that, even for large data sets. Thanks to the scale-up and scale-out features of Ehcache, most users can keep 100% of their data in cache, not just 20%.