Neal, D. The Harvest object cache in New Zealand. Proc. Of the 5 World Wide Web Conference Paris, France, May 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....provide them to other clients usenj more quickly than can the originating Web sj7 er. In dis8UjMP;9 Web caching, each cachesj7 er provides Web objects to the other cachesj7 ers optimizing the retrieval of cached Web objects Dis;jMP7S cachesjR ers are currently usj to provide caching for the Web [2], 3] A disUSjMLLU cachesj7 erdes87LU on a large sgej sgejP reduce retrieval latency further through much more e#cientreus8 Many disSjM9;U; cachesjP er shemes have been Manuscript receivedOct5 er 15, 1998. Manuscript revised January 8, 1999. Theaut911 are wit NTT Service IntL1L Laborat958 ....

D. Neal,"The Harvest object cache in New Zealand," Compu -, Networks and ISDN Systems,vol.28,no.6,p.14151430, May 1996.

....documents to satisfy subsequent access to the same document. At the server side [6, 7, 8, 9, 18] recently requested documents can be stored in memory to satisfy future requests to the same document by the same user or another user, thus eliminating disk I O access time. Within the network [1, 10, 12, 13, 18, 20, 28, 56, 59, 62, 65, 74, 76], proxy caches can be employed to act as intermediaries between Web clients (browsers) and Web servers, accepting user requests and forwarding them to Web servers only as necessary. When a requested document is returned by a Web server, the proxy server sends the document to the requesting client ....

....the national cache. If the request is not found at any cache level, the national cache contacts the origin server. When the document is found, either at a cache or at the origin server, it travels down the hierarchy, leaving a copy at each of the intermediate caches. Squid [52, 65, 70] and HARVEST [12, 27, 56, 70] are among the commonly used proxy software programs. 2.6 Related Work The explosive growth of the Web has been the main reason for increasing research attention on the Web. Improving the scalability of the Web is much dependent on the components that make up the Web the clients, the servers, ....

[Article contains additional citation context not shown here]

D. Neal, \The Harvest Object Cache in New Zealand," in Proceedings of the 5th International World-Wide Web Conference, May 1996.

....important issues, such as server loads, and client response times. In a more recent and exhaustive study, Cao and Liu [22] argue that strong cache consistency can be maintained with little or no extra cost. 2.2. 4 Web Cache Hierarchies Although caching is increasingly being used in the Internet [58, 76] to reduce network trac and distribute load away from server hot spots , it has been less successful in reducing access latencies [80] Cache hits at the higher levels require the document to be sent down the hierarchy, with each cache along the path storing a copy of the document being fetched. ....

D. Neal, \The HARVEST Object Cache in New Zealand", Computer Networks and ISDN Systems, Vol. 28, No. 7-11, pp. 1415-1430, May 1996. 119

....server, variable size Figure 5: Hit rates of various total cache sizes under HRW 2.8 3 3.2 3.4 3.6 3.8 4 4. 2 1 2 3 4 5 6 7 8 Average Latency (sec) # Servers HRW LeastJobs Random Round Robin Figure 6: Latency of various allocation schemes the outside world, such as New Zealand [30], since MISS will rise while HIT remains constant. Figure 8 shows the hit rate and cache space used when no limit exists on cache space. Again, since we assume that no objects are invalidated during the lifetime of the simulation, no time based expirations were simulated, and hence no objects ....

Donald Neal. The Harvest object cache in New Zealand. In Fifth International World Wide Web Conference, May 1996.

....obtained from five sources. Two were downloaded from public FTP sites at Boston University [5] and Virginia Tech. 6] Another two were provided by the Information Technology Services at the University of Waikato which administers the cache installed on the New Zealand Internet Exchange (NZIX) [7] as well as the Campus Cache of the University of Waikato. The remaining two workloads were collected from the cache of the University of Waikato Computer Science Department. The six workloads are: Boston University (BU) Virginia Tech (VT) New Zealand Internet Exchange (NZIX) The University of ....

D Neal, The Harvest Object Cache in New Zealand, Proceedings of the Fifth International World Wide Web Conference, Paris, France, May 1996.

....to provide improved performance for all of their clients. We believe our model will be interesting to system designers building large scale, distributed cache infrastructures in a range of environments including network service providers, independent service providers, cache service providers [24, 85, 108, 123], collections of caches linked by formal service agreements [107] and large intranets. Also, once the trade offs in these cooperative environments are understood, we plan to examine ways to apply some of these techniques in more competitive environments. The rest of the chapter is organized as ....

....server. We reach similar conclusions in the context of Internet caching, leading to our design principle of minimizing the number of hops on a hit or miss. Hierarchical caching has been more widely accepted on the Internet, and it is common in areas with low speed connections to the Internet [85, 108]. In addition, many sites add proxy caching to their corporate firewalls. A widely deployed and studied system has been the publicly available Harvest cache [19] and its successor, the Squid cache [123] A key motivation of our design is that we disagree with the conclusion that cache hierarchies ....

D. Neal. The Harvest Object Cache in New Zealand. In Proceedings of the Fifth International World Wide Web Conference, May 1996.

.... of Colorado and the University of Southern California developed a hierarchical caching strategy [3] as part of the Harvest resource discovery project[11] The Harvest cache (and a later version named Squid ) is currently being used to provide caching for the World Wide Web in New Zealand [8], and as the infrastructure for a distributed information provision testbed developed by the National Laboratory for Applied Network Research (NLANR) 7] The goals of the hierarchical cache are to distribute load away from server hot spots and to reduce access latency. It aims to do this by ....

Donald Neal. The Harvest object cache in New Zealand. Available from http://- www.waikato.ac.nz/harvest/www5/- Overview.html, 1996.

....[1] This demand for data naturally creates network congestion and tends to increase retrieval latency. In order to reduce the demand for bandwidth and to improve retrieval latency, several agencies have begun to implement network caching for the internet (examples may be found in [2] 3] and [4]) The main idea in network caching is to create copies of network accessible documents which are closer (either physically, with respect to transmission cost, or with respect to latency) than the home location of the document. The current efforts have largely been based on demand placement ....

Donald Neal, "The Harvest Object Cache in New Zealand," Fifth International World Wide Web Conference, http://www5conf.inria.fr/fich_html/papers/P46/Overview.html, May 6-10, 1996.

....to provide improved performance for all of their clients. We believe our model will be interesting to system designers building largescale, distributed cache infrastructures in a range of environments including network service providers, independent service providers, cache service providers [6, 29, 35, 39], collections of caches linked by formal service agreements [34] and large intranets. Also, once the trade offs in these cooperative environments are understood, we plan to examine ways to apply some of these techniques in more competitive environments. The rest of the paper is organized as ....

....server. We reach similar conclusions in the context of Internet caching, leading to our design principle of minimizing the number of hops on a hit or miss. Hierarchical caching has been more widely accepted on the Internet, and it is common in areas with low speed connections to the Internet [29, 35]. In addition, many sites add proxy caching to their corporate firewalls. A widely deployed and studied system has been the publicly available Harvest cache [5] and its successor, the Squid cache [39] A key motivation of our design is that we disagree with the conclusion that cache hierarchies ....

D. Neal. The Harvest Object Cache in New Zealand. In Proceedings of the Fifth International World Wide Web Conference, May 1996.

....and possibly cached, in case they are requested again. Proxy servers provide several benefits, such as reduced web latency and increased effective bandwidth[12] increased web access (when the destination server is down, but the resource is cached) and savings on long distance connection charges[15]. The two main problems for a proxy server implementation are how to determine if a cached file is fresh and which files to remove when the cache gets full. Freshness is difficult to determine because few servers use the Expires HTTP Fig. 3. Snapshot of a Dynamic Cache Visualization. header and ....

Donald Neal. The harvest object cache in New Zealand. In Proceedings of the Fifth International World Wide Web Conference, 1996. http://www5conf.inria.fr/fich html/papers/P46/Overview.html.

....of reference properties for the WWW access patterns are not strong enough to result in an effective caching strategy. However at much larger scale these properties could be exploit and that is what better client schemes try to do. A real implementation of a proxy cache system is the CERN software [2]. A CERN proxy server may be also configured to use another server as its proxy, referring requests not satisfied by hits on its own cache to that other server. A tree of servers may be constructed, offering greater savings that only one proxy server alone. The main weakness of this system is that ....

D. Neal, "The harvest object cache in new zealand," in Computer Networks and ISDN Systems, (Hamilton, New Zealand), 1996.

....of caches with tens of thousands to millions of users. We believe our techniques will be of interest to system designers building large scale, distributed cache infrastructures in a range of environments including network service providers, independent service providers, cache service providers [9, 33, 40, 45], collections of caches linked by formal service agreements [39] and large intra nets. Using measurements of several caches on the Internet and analysis of several traces of web traffic, we first attempt to understand the factors that limit the performance of current web caches. We find that to ....

D. Neal. The Harvest Object Cache in New Zealand. In Proceedings of the Fifth International World Wide Web Conference, May 1996.

....a thousand OC3 circuits, but downloading the latest version of MSIE 1 will still take 20 hours for some folks in Moscow who have a single 19.2 kbps link for all of their simultaneous TCP IP connections. Caching has proven a useful technique for reducing end user experienced latency on the Web [1, 2, 3, 4, 5, 6, 7]. The fundamental concept is the intermediate storage of copies of popular Web documents close to the end users. Caching is effective because many Web documents are requested (much) more than once [8] Web browsers have local disk caches because individuals often browse the same pages repeatedly. ....

D. Neal, "The harvest object cache in new zealand," in Proccedings of the World Wide Web Conference, May 1995.

No context found.

Neal, D. The Harvest object cache in New Zealand. Proc. Of the 5 World Wide Web Conference Paris, France, May 1996.

No context found.

D. Neal, "The Harvest object cache in New Zealand.," Computer Networks and ISDN Systems, vol. 28, no. 7-11, pp. 1415--30, May 1996, (Fifth International World Wide Web Conference, Paris, France, 6-10 May 1996.).

No context found.

D. Neal, "The Harvest object cache in New Zealand.," Computer Networks and ISDN Systems, vol. 28, no. 7-11, pp. 1415--30, May 1996, (Fifth International World Wide Web Conference, Paris, France, 6-10 May 1996.).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC