Amortized efficiency of list update and paging rules
Communications of the ACM
Journal of Algorithms
Page replacement with multi-size pages and applications to Web caching
STOC '97 Proceedings of the twenty-ninth annual ACM symposium on Theory of computing
STOC '99 Proceedings of the thirty-first annual ACM symposium on Theory of computing
Proceedings of the ninth annual ACM-SIAM symposium on Discrete algorithms
Competitive analysis of randomized paging algorithms
Theoretical Computer Science
Connection caching under various models of communication
Proceedings of the twelfth annual ACM symposium on Parallel algorithms and architectures
Generalized connection caching
Proceedings of the twelfth annual ACM symposium on Parallel algorithms and architectures
Web caching with request reordering
SODA '02 Proceedings of the thirteenth annual ACM-SIAM symposium on Discrete algorithms
SODA '02 Proceedings of the thirteenth annual ACM-SIAM symposium on Discrete algorithms
Paging with connections: FIFO strikes again
Theoretical Computer Science
Hi-index | 0.00 |
Document caching and connection caching are extensively studied problems. In document caching, one has to maintain caches containing documents accessible in a network. In connection caching, one has to maintain a set of open network connections that handle data transfer. Previous work investigated these two problems separately while in practice the problems occur together: In order to load a document, one has to establish a connection between network nodes if the required connection is not already open. In this paper we present the first study that integrates document and connection caching. We first consider a very basic model in which all documents have the same size and the cost of loading a document or establishing a connection is equal to 1. We present deterministic and randomized online algorithms that achieve nearly optimal competitive ratios unless the size of the connection cache is extremely small. We then consider general settings where documents have varying sizes. We investigate a FAULT model in which the loading cost of a document is 1 as well as a BIT model in which the loading cost is equal to the size of the document.