Performance Analysis Methods for List-Based Caches With Non-Uniform Access

List-based caches can offer lower miss rates than single-list caches, but their analysis is challenging due to state space explosion. In this setting, we propose novel methods analyze performance for a general class of list-based with tree structure, non-uniform access items and lists, random or first-in first-out replacement policies. Even though the underlying Markov process shown admit product-form solution, difficult exploit large caches. Thus, develop approximations cache metrics, in particular by means singular perturbation method refined mean field approximation. We compare accuracy these approaches simulations, finding that our new rapidly converge equilibrium distribution as number capacity grow fixed ratio. find they are much more accurate point similar prior work, average errors typically below 1.5% even very small Our models also generalized account synchronous requests, fetch latency, item sizes, extending applicability