[ Homepage  | Biography  | Courses  | Members  | Publications  | Research/Funding  | Workshops/Presentations  ]

Cache Replacement Algorithms for Large Data Objects (Video)

Partial caching has been recently introduced as a new idea for the caching of large data objects such as video files. The motivation for partial caching has been that the caching of entire large objects can easily exhaust the storage resources of a proxy server. Our research combines partial caching with demand driven replacement algorithms and explores the impact of replacement granularity on the caching of large media objects. In particular the idea of segmenting video files into chunks and applying replacement decisions at the chunk level rather than on entire videos is examined. In our works it is shown that a higher byte hit ratio (BHR) can be achieved by appropriately adjusting the replacement granularity. The price paid for the improved BHR performance is that the replacement algorithm takes a longer time to converge to the steady state BHR. Several segmentation schemes have been studied and evaluated with respect to the steady state BHR and convergence time. It is also shown that by properly segmenting video objects, caching differentiation based on access costs can be achieved.

This line of work is part of our more general involvement in the study of content distibution in the internet. Other related activities include: replacement algorithms and segmentation of objects in hierarchical caches as well as applications of caching for mobile users.

The produced results have been recorded in the following publications: