Understanding cache metrics

Cache hit ratio metric information intrinsically depends on the workflow that is using the object. Some workflows involve reading an object only one time while others involve reading the object many times. The cache hit varies depending on these workflows.

Also, it is also for the data be appear skewed to the low side if a server started recently, or if the server has not had much user load. For example, if you recently started the server, and users have only visited a few screens, the hit rate is very low because PolicyCenter encountered only a few cache hits. As users visit more pages, the hit rate increases.

Thus, there are no good default cache hit ratios. Experimentation combined with performance measurements constitutes the only approach to identifying appropriate cache sizes.