The Case for Market-Based Push Caching

Yee Man Chan, Johnathan Womer, Sugih Jamin, Jeffrey K. MacKie-Mason
1999 Social Science Research Network  
Web caches currently deployed on the Internet operate under a pull model in which client request streams determine the content of the cache. An alternative push model would allow web servers to pro-actively replicate their contents to caches. Given the finite amount of cache space, a question arises as to which objects should be kept in cache. In [6] , the authors propose a push caching model whereby the content of the cache is determined by participating servers in a cooperative fashion. In
more » ... s paper we explore an auction-based scheme that achieves an efficient allocation of disk space based on utilities revealed by both content ¤ This project is funded in part by DARPA/ITO grant F30602-97-1-0228 from the Information Survivability program. ¥ Yee Man Chan appreciates the equipment and financial support from EnjoyWeb, Inc. ¦ Sugih Jamin is supported in part by NSF CAREER Award ANI-9734145, Presidential Early Career Award for Scientists and Engineers (PECASE) 1998, and equipment support from Sun Microsystems Inc. and Compaq (Digital Equipment) Corp. § MacKie-Mason gratefully acknowledges the support of an IBM University Partnership grant. servers and web caches. We show that our approach provides higher user valuation than traditional replacement policies without sacrificing overall hit rate. At the same time we solve the truthful revelation incentive problems associated with a cooperative approach. Our approach both implements e-commerce in caching service, and improves the infrastructure for supporting other e-commerce by providing quality of service differentiation.
doi:10.2139/ssrn.1172462 fatcat:vhd5273whraadfq33mwak6xn2a