Minimization of Download Times for Large Files over Wireless Channels
IEEE Transactions on Mobile Computing
GPC'12 Proceedings of the 7th international conference on Advances in Grid and Pervasive Computing
| Hi-index | 0.00 |
Numerous research efforts have produced a largenumber of algorithms and mechanisms for web proxycaches. In order to build powerful web proxies andunderstand their performance, one must be able toappreciate the impact and significance of earliercontributions and how they can be integrated. To do thiswe employ a cache replacement algorithm, 'CSP', whichintegrates key knowledge from previous work. CSPutilizes the communication Cost to fetch web objects, theobjects' Sizes, their Popularities, an auxiliary cache anda cache admission control algorithm. We study theimpact of these components with respect to hit ratio,latency, and bandwidth requirements.Our results show that there are clear performancegains when utilizing the communication cost, thepopularity of objects, and the auxiliary cache. Incontrast, the size of objects and the admission controllerhave a negligible performance impact. Our majorconclusions going against those in related work are that(i) LRU is preferable to CSP for important parametervalues, (ii) accounting for the objects' sizes does notimprove latency and/or bandwidth requirements, and(iii) the collaboration of nearby proxies is not verybeneficial. Based on these results, we chart the problemsolution space, identifying which algorithm is preferableand under which conditions. Finally, we develop adynamic replacement algorithm that continuouslyutilizes the best algorithm as the problem-parametervalues (e.g., the access distributions) change with time.