Desenvolvimento de um simulador de cache para análise de políticas de gerenciamento.

Abstract

Caching is a traditional strategy for improving the performance of computing systems. Cache systems have several settings that can change their operation and consequently their performance. This is the case, for example, with the parameterization of item removal policies. When the scale of the cache system becomes very large, experiments to test alternatives to these configurations become difficult. Considering this problem, we developed a simulator that models a popular caching system, NGINX. We validated this simulator using traces of a cache deployment for a high-scale web system. The validation of our simulator indicates that it presents low error, around 1% of the requests.