New adaptive object replacement policy for video-on-demand systems

摘要

The architecture of the Berkeley distributed hierarchicalnvideo-on-demand (VOD) system consists of continuous media servers (CMS)nbasically standing as a cache between the archive servers (AS) and thenclients. Many replacement policies used in cache management, we notice,nis restricted in that they consider only one (or few) parameters tondecide which video to kick out of the cache, like the frequency ofnrequests, or the size of objects. We look at a new cache managementnalgorithm which is higher-performing since it considers, besides thentime of requests and the past request history of a video, the videoncharacteristics which viewers, generally, have a bias for. However, thenviewer's bias and the objects' characteristics vary with time andnlocation. We enhanced the cache management technique by making itnadaptive to the different characteristics viewers and videos have atndifferent times and places