Estimating Cache Hit Rates From the Miss Sequence
Click here to download now
Overview: This paper published by HP considers the problem of inferring cache hit rates from observations of references that miss in the cache. In the Web context, this is the problem of inferring a browsers cache hit rate by examining the requests that it issues to origin servers. The paper shows that a case of this problem can be solved. Specifically it considers a sequence of references to unit-sized objects generated by an LRU (Least Recently Used) stack distance model and filtered by an LRU cache of known size.