## On-line file caching (1998)

### Cached

### Download Links

- [www.cs.ucr.edu]
- [www.cs.ucr.edu]
- [www.drenchedrock.net]
- [ftp.cs.dartmouth.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | In Proc. of the 9th Annual ACM-SIAM Symp. on Discrete algorithms |

Citations: | 75 - 2 self |

### BibTeX

@INPROCEEDINGS{Young98on-linefile,

author = {Neal E. Young},

title = {On-line file caching},

booktitle = {In Proc. of the 9th Annual ACM-SIAM Symp. on Discrete algorithms},

year = {1998}

}

### Years of Citing Articles

### OpenURL

### Abstract

Consider the following file caching problem: in response to a sequence of requests for files, where each file has a specified size and retrieval cost, maintain a cache of files of total size at most some specified k so as to minimize the total retrieval cost. Specifically, when a requested file is not in the cache, bring it into the cache, pay the retrieval cost, and choose files to remove from the cache so that the total size of files in the cache is at most k. This problem generalizes previous paging and caching problems by allowing objects of arbitrary size and cost, both important attributes when caching files for world-wide-web browsers, servers, and proxies. We give a simple deterministic on-line algorithm that generalizes many well-known paging and weighted-caching strategies, including least-recently-used, first-in-first-out,

### Citations

754 |
Amortized efficiency of list update and paging rules
- SLEATOR, TARJAN
- 1985
(Show Context)
Citation Context ...g it into the cache, paying the retrieval cost of the file, and choose files to remove from the cache so that the total size of files remaining in the cache is at most k. Following Sleator and Tarjan =-=[15]-=-, we say a file caching algorithm is c(h; k)-competitive if on any sequence the total retrieval cost incurred by the algorithm using a cache of size k is at most c(h; k) times the minimum possible cos... |

581 | Consistent hashing and random trees: Distributed caching protocols for relieving hot spots on the World Wide Web
- Karger, Lehman, et al.
- 1997
(Show Context)
Citation Context ...n about the change in some way). Finally, the focus here is on simple local caching strategies, rather than distributed strategies in which servers cooperate to cache pages across a network (see e.g. =-=[9]-=-). 4 This performance guarantee is the best possible for any deterministic on-line algorithm. 2 File caching is not a special case of the k-server problem, although weighted caching is a special case ... |

495 | Cost-aware www proxy caching algorithms
- Cao, Irani
- 1997
(Show Context)
Citation Context ...alysis uses the potential function = (h 1) P f2ll credit[f ]+k P f2opt cost(f) credit[f ]. The analysis is simpler than that of [18] for the special case of weighted caching. In an independent work [=-=2]-=-, Cao and Irani showed that Landlord (with step 7 raising credit[g] as much as possible) is k-competitive. They also gave empirical evidence that the algorithm performs well in practice. This paper: (... |

167 | Competitive paging algorithms
- Fiat, Karp, et al.
- 1991
(Show Context)
Citation Context ...performance guarantee is the best possible for any deterministic on-line algorithm. A simple randomized paging algorithm called the marking algorithm was shown to be 2 ln k-competitive by Fiat et al. =-=[5]-=-. An optimal ln k-competitive randomized paging algorithm was given by McGeoch and Sleator [14]. In [18], deterministic paging strategies were shown to be loosely O(ln k)-competitive. This means rough... |

142 |
Competitive algorithms for server problems
- Manasse, McGeoch, et al.
- 1990
(Show Context)
Citation Context ...tion time. Similar considerations hold in other settings such assle caching. To formalize this, we introduce a parameter > 0, and say that \low enough" for a request 2 Manasse, McGeoch, and Slea=-=tor [13]-=- show that no deterministic on-line algorithm for the well-known k-server problem on any metric space of more than k points is better than k k h+1 - competitive. This implies that, at least for any sp... |

124 | Competitive paging with locality of reference
- Borodin, Irani, et al.
- 1995
(Show Context)
Citation Context ...o the theoretically optimal competitive ratio of k. A number of renements of competitive analysis have been proposed to try to understand the relevant factors. Borodin, Irani, Raghavan, and Schieber [=-=1]-=-, in order to model locality of reference, proposed the access-graph model which restricts the request sequences to paths in a given graph (related papers include [4, 8, 6]). Karlin, Phillips, and Rag... |

123 | Beyond competitive analysis
- Koutsoupias, Papadimitriou
(Show Context)
Citation Context ...ant in which the graph is a Markov chain (i.e. the edges of the graph are assigned probabilities, and the request sequence corresponds to a random walk) (see also [12]). Koutsoupias and Papadimitriou =-=[11-=-] proposed the comparative ratio (for comparing classes of on-line algorithms) and the diuse adversary model (in which the adversary chooses a probability distribution, rather than a sequence, from so... |

115 |
A strongly competitive randomized paging algorithm
- McGeoch, Sleator
- 1991
(Show Context)
Citation Context ...andomized paging algorithm called the marking algorithm was shown to be 2 ln k-competitive by Fiat et al. [5]. An optimal ln k-competitive randomized paging algorithm was given by McGeoch and Sleator =-=[14]-=-. In [18], deterministic paging strategies were shown to be loosely O(ln k)-competitive. This means roughly that for any sequence, for most values of k, the fault rate of the algorithm using a cache o... |

75 |
Tarjan. Amortized eciency of list update and paging rules
- Sleator, E
- 1985
(Show Context)
Citation Context ...ache, bring it into the cache, paying the retrieval cost of thesle, and remove othersles from the cache so that the total size ofsles remaining in the cache is at most k. Following Sleator and Tarjan =-=[15]-=-, we say asle caching algorithm is c(h; k)- competitive if on any sequence the total retrieval cost incurred by the algorithm using a cache of size k is at most c(h; k) times the minimum possible cost... |

74 | New results on server problems
- Chrobak, Karloff, et al.
- 1990
(Show Context)
Citation Context ...)- competitive. Uniform sizes, arbitrary costs. The special case ofsle caching when allsle sizes are the same is called weighted caching. For weighted caching, Chrobak, Karlo, Payne and Vishwanathan [=-=3] showed th-=-at an algorithm called the \balance" algorithm is k-competitive. Subsequently in [18] a generalization of that algorithm called the \greedy-dual" algorithm was shown to be k k h+1 - competit... |

74 | Strongly competitive algorithms for paging with locality of reference
- IRANI, KARLIN, et al.
- 1996
(Show Context)
Citation Context ...in, Irani, Raghavan, and Schieber [1], in order to model locality of reference, proposed the access-graph model which restricts the request sequences to paths in a given graph (related papers include =-=[4, 8, 6]-=-). Karlin, Phillips, and Raghavan [10] proposed a variant in which the graph is a Markov chain (i.e. the edges of the graph are assigned probabilities, and the request sequence corresponds to a random... |

63 | The k-server dual and loose competitiveness for paging
- Young
- 1994
(Show Context)
Citation Context ...d paging algorithm called the marking algorithm was shown to be 2 ln kcompetitive by Fiat et al. [4]. An optimal ln k-competitive randomized paging algorithm was given by McGeoch and Sleator [14]. In =-=[18]-=-, deterministic paging strategies were shown to be loosely O(ln k)-competitive. This means roughly that for any sequence, for most values of k, the fault rate of the algorithm using a cache of size k ... |

60 |
Replacement with Multi-Size Pages and Applications to Web Caching
- Page
(Show Context)
Citation Context ...re either all equal (the goal is to minimize the number of retrievals), and when each cost equals thesle size (the goal is to minimize the total number of bytes retrieved). For these two cases, Irani =-=[7]-=- gave O(log 2 k)-competitive randomized on-line algorithms. Comment: the importance of sizes and costs. File caching is important for world-wide-web applications. For instance, in browsers and proxy s... |

46 |
On-line caching as cache size varies
- Young
- 1991
(Show Context)
Citation Context ...ider randomized algorithms: Theorem 3 Let 0 ;s 1. Any +sln k k h+1 -competitive algorithm is (; )-loosely c-competitive for c = e+esln[(1=) ln(e=)] = O(log[(1=) log(1=)]). It is known (e.g. [17, 16]=-=-=-) that the marking algorithm (a randomized on-line algorithm) is (1+2 ln k k h )-competitive for paging and (1+2 ln k)-competitive for h = k. It follows by algebra that the marking algorithm is 1+2 ln... |

38 |
Randomized and Multipointer paging with locality of reference
- Fiat, Karlin
- 1995
(Show Context)
Citation Context ...in, Irani, Raghavan, and Schieber [1], in order to model locality of reference, proposed the access-graph model which restricts the request sequences to paths in a given graph (related papers include =-=[4, 8, 6]-=-). Karlin, Phillips, and Raghavan [10] proposed a variant in which the graph is a Markov chain (i.e. the edges of the graph are assigned probabilities, and the request sequence corresponds to a random... |

31 | Experimental studies of access graph based heuristics. Beating the LRU standard
- Fiat, Rosen
- 1997
(Show Context)
Citation Context ...in, Irani, Raghavan, and Schieber [1], in order to model locality of reference, proposed the access-graph model which restricts the request sequences to paths in a given graph (related papers include =-=[4, 8, 6]-=-). Karlin, Phillips, and Raghavan [10] proposed a variant in which the graph is a Markov chain (i.e. the edges of the graph are assigned probabilities, and the request sequence corresponds to a random... |

20 |
IP over connection-oriented networks and distributional paging
- Lund, Phillips, et al.
- 1994
(Show Context)
Citation Context ...ps, and Raghavan [10] proposed a variant in which the graph is a Markov chain (i.e. the edges of the graph are assigned probabilities, and the request sequence corresponds to a random walk) (see also =-=[12-=-]). Koutsoupias and Papadimitriou [11] proposed the comparative ratio (for comparing classes of on-line algorithms) and the diuse adversary model (in which the adversary chooses a probability distribu... |

17 |
The k-server dual and loose competitiveness for paging. Algorithmica,June 1994, vol. 11,(no.6):525-41. Rewritten version of \Online caching as cache size varies
- Young
- 1991
(Show Context)
Citation Context ... paging algorithm called the marking algorithm was shown to be 2 ln k-competitive by Fiat et al. [5]. An optimal ln k-competitive randomized paging algorithm was given by McGeoch and Sleator [14]. In =-=[18]-=-, deterministic paging strategies were shown to be loosely O(ln k)-competitive. This means roughly that for any sequence, for most values of k, the fault rate of the algorithm using a cache of size k ... |

8 |
Competitive paging and dual-guided algorithms for weighted caching and matching
- Young
- 1991
(Show Context)
Citation Context ...ider randomized algorithms: Theorem 3 Let 0 ;s 1. Any +sln k k h+1 -competitive algorithm is (; )-loosely c-competitive for c = e+esln[(1=) ln(e=)] = O(log[(1=) log(1=)]). It is known (e.g. [17, 16]=-=-=-) that the marking algorithm (a randomized on-line algorithm) is (1+2 ln k k h )-competitive for paging and (1+2 ln k)-competitive for h = k. It follows by algebra that the marking algorithm is 1+2 ln... |

4 |
Markov paging (extended abstract
- Karlin, Phillips, et al.
- 1992
(Show Context)
Citation Context ...rder to model locality of reference, proposed the access-graph model which restricts the request sequences to paths in a given graph (related papers include [4, 8, 6]). Karlin, Phillips, and Raghavan =-=[10]-=- proposed a variant in which the graph is a Markov chain (i.e. the edges of the graph are assigned probabilities, and the request sequence corresponds to a random walk) (see also [12]). Koutsoupias an... |

4 |
McGeoch and Daniel Dominic Sleator. A strongly competitive randomized paging algorithm
- Lyle
- 1991
(Show Context)
Citation Context ...randomized paging algorithm called the marking algorithm was shown to be 2 ln kcompetitive by Fiat et al. [4]. An optimal ln k-competitive randomized paging algorithm was given by McGeoch and Sleator =-=[14]-=-. In [18], deterministic paging strategies were shown to be loosely O(ln k)-competitive. This means roughly that for any sequence, for most values of k, the fault rate of the algorithm using a cache o... |

3 | 35th Annual Symposium on Foundations of Computer Science, Sant,a Fe - IEEE - 1994 |