Efficiency of cache-replacement algorithms while retrieving data from a relational database and XML files in a web based system

Abstract

Caching has been applied in Web based information systems in order toreduce the transmission of redundant network traffic and response latency by savingcopies of the content obtained from the Web closer to the end user. The efficiencyof caching is influenced to a significant extent by the cache replacement algorithmswhich are triggered when the cache becomes full and old objects must be evicted tomake space for the new ones.This paper presents a framework that can be used in future work to tunecache-replacement algorithms while data is simultaneously retrieved from arelational database and XML files in a web based environment, by a large numberof end-users. Three replacement policies are considered: Least Recently Used(LRU), Least Frequently Used (LFU) and Lowest Latency First (LLF). Theexperimental results obtained from the framework show that data caching greatlyimproves the overall performance of web based systems, and the type of the appliedcache replacement policy also plays an important role in the performance. In thescenarios considered in this paper, the LLF algorithm produced the bestperformance when retrieving data from a relational database, while the LFUalgorithm was the most efficient algorithm when data was retrieved from an XMLfile.