News‎ > ‎

Paper on "A New Class of Web Caching Strategies for Content Delivery" accepted at NETWORKS 2014

posted Jul 22, 2014, 4:39 AM by Patrick Poullie   [ updated Jul 22, 2014, 4:40 AM ]
A paper on "A New Class of Web Caching Strategies for Content Delivery" by Gerhard Hasslinger, Kostas Ntougias and Frank Hasslinger has been accepted at the NETWORKS 2014: 16th International Telecommunications Network Strategy and Planning Symposium to be held in Sept. 17-19 in Funchal, Madeira, Portugal. The Networks Symposium is a biannual event mainly organized by network providers and telecommunications industry together with academia. Our contribution addresses efficient caching methods to be applied in CDNs, local and ISP networks, which are relevant as a basic function in SmartenIT components.

Abstract of the paper:

Least recently used (LRU) is the most commonly applied strategy to update the content of caches in computing and database systems as well as for web caching. Although some case studies have shown that LRU hit rates for web caching can be low when compared to strategies being aware of complete statistics of past requests, proposed alternatives seem too complex to cope with constant update effort of LRU.

In this work we start with an exhaustive evaluation of the hit rates for LRU as compared to statistic-based caching strategies for Zipf distributed content popularity, which has been confirmed manifold as the relevant access profile to content on the Internet. We conclude that LRU has more than 10% absolute hit rate deficits not only in some special cases but over the entire relevant parameter range of Zipf law access pattern.

On the other hand, we show that a 10% improvement over LRU hit rates is already realized by the variant of score-gated LRU, whose mean updating effort is comparable to pure LRU. Furthermore, score-gated LRU avoids most of the input traffic of a pure LRU strategy, which frequently reloads the same objects into the cache. Score-gated LRU keeps the cache content stable in case of unchanged popularity and loads new objects only when their score is increasing and exceeds the score of a cached object.