A study of the performance and parameter sensitivity of adaptive distributed caching

A self-organized approach to manage a distributed proxy system called Adaptive Distributed Caching (ADC) has been proposed previously [8]. We model each proxy as an autonomous agent that is equipped to decide how to deal with client requests using local information. Experimental results show that our ADC algorithm is able to compete with typical hashing-based approaches. This paper gives a full description of the self-organizing distributed algorithm, with a performance comparison based on hits and hops rate. Additional evaluation of the parameters caching-, multiple- and single-table size is also presented.