Wang, Z., Crowcroft, J., "Cachemesh: A Distributed Cache System For World Wide Web," NLANR Web Cache Wk., Boulder, Jun. 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and content requested, networks such as CDNs are bound to grow (in terms of the number of proxies for instance) to meet demands. Hence there is a need to develop scalable and efficient solutions for cache consistency to deal with such growths. While studies have addressed viable CDN architectures [48, 8, 4, 33, 12, 43, 37, 16], the role of a proxy in a CDN [32] cooperative caching to improve hit ratio and response times [44, 5, 11, 43, 12, 17, 28, 34, 38, 49] load balancing amongst proxies [23, 24, 33] redirection schemes and other performance issues [21, 20, 31] object replication strategies in such networks [22] ....

....Hence there is a need to develop scalable and efficient solutions for cache consistency to deal with such growths. While studies have addressed viable CDN architectures [48, 8, 4, 33, 12, 43, 37, 16] the role of a proxy in a CDN [32] cooperative caching to improve hit ratio and response times [44, 5, 11, 43, 12, 17, 28, 34, 38, 49], load balancing amongst proxies [23, 24, 33] redirection schemes and other performance issues [21, 20, 31] object replication strategies in such networks [22] and how prefetching affects performance in CDNs [42] very few have addressed consistency issues in CDNs. 47, 46, 48, 45, 33] are few ....

[Article contains additional citation context not shown here]

Z. Wang. Cachemesh: A distributed cache system for world wide web, 1997.

....proxy cache [10] They lead to overhead problems like inter proxy communication and the distribution of object storage location information. Previous research on cooperative proxies can be found in hierarchical [33] and hashing approaches [20, 29] adaptive web caching [37] CacheMesh [34], WebWave [15] and the straightforward approach with a central coordinator [25] Additional research for distributed systems covers areas on consistency between multiple proxies [14] the theoretical description of the underlying processes [21, 35] and attempts to build the system on the idea of ....

....or requested from the origin server. Hashing based allocations can be widely seen 4 TSUI et al. as the ideal way to find cached web pages, due to the fact that their location is pre defined. Their major drawback is inflexibility and poor adaptability. Adaptive Web Caching [37] and CacheMesh [34] try to overcome specific performance bottlenecks. For example, Adaptive Web Caching dynamically creates proxy groups combined with data multicasting, while CacheMesh computes the routing protocol based on exchanged routing information. Both approaches can still be considered experimental. Yet ....

Z. Wang and J. Crowcroft. CacheMesh: A distributed cache system for world wide web. In Proceedings of NLANR Web Cache Workshop, June 1997.

....function to route requests directly to one of a number of loosely coupled proxies. The CRISP web caching service [14] proposed to use a central ized directory for each cluster of proxies. A user s proxy would contact the directory to find of copy of the cached content. The Cachemesh project [41] pro posed to use communication among proxies to develop cache placement and routing strategies. Cache Digests [34] and the Summary Cache [12] use Bloom filters to store a summary of the directory at each node. Proxies consult the bloom filters stored locally to find another proxy which (with ....

Z. Wang and J. Crowcroft. Cachemesh: a distributed cache system for the world wide web. In Web Cache Workshop, 1997.

....hashing function to route requests directly to one of a number of loosely coupled proxies. The CRISP web caching service [14] proposed to use a centralized directory for each cluster of proxies. A user s proxy would contact the directory to nd of copy of the cached content. The Cachemesh project [41] proposed to use communication among proxies to develop cache placement and routing strategies. Cache Digests [34] and the Summary Cache [12] use Bloom lters to store a summary of the directory at each node. Proxies consult the bloom lters stored locally to nd another proxy which (with high ....

Z. Wang and J. Crowcroft. Cachemesh: a distributed cache system for the world wide web. In Web Cache Workshop, 1997.

....on web servers can be reduced by distributed web server architectures that distribute client requests among multiple geographically dispersed server nodes in a user transparent way [29] Another approach is to redirect requests to web caches. For example, in the distributed cache system Cachemesh [30] each cache server becomes the designated cache for several web sites. Requests for objects not in the cache are forwarded to the responsible cache. However, not all web data is cacheable, in particular dynamic and personalized data. Also, load balancing mechanisms on the front ends of web server ....

Wang Z.: Cachemesh: A Distributed Cache System for the World Wide Web. 2nd NLANR Web Caching Workshop (1997)

.... their individual caches in such a way that maximum cache usage is achieved while acting transparently as one single loadbalanced proxy cache [11] Previous research on cooperative proxies can be found in the area of hierarchical [7] and hashing approaches [9] adaptive caching [2] CacheMesh [1] and our previously introduced algorithms SOAP [5] and ADC [3] 1.1. Hierarchical vs. Hashing Common approaches for cooperative proxy systems encompass mostly solutions based on hierarchical cache structures or classical hashing algorithms [9] While hierarchically organized proxies forward ....

.... we order the mapping table based on the average request time, an object that has been requested only two times but within a very short time frame and never again will be wrongly placed into the mapping table (or even the caching part in the worst case) www.xy34 Proxy[4] 9953 0 1 www.xy123 Proxy[1] 9954 0 1 www.xy64 Proxy[2] 9955 0 1 www.xy53 Proxy[1] 9956 123 432 www.xy343 Proxy[7] 9957 0 1 www.xy29 Proxy[4] 9961 0 1 Figure 3. A Sample Single Table www.xy55 This 4253 75 2 www.xy13 Proxy[1] 4154 83 34 www.xy644 This 6555 90 2 www.xy52 Proxy[4] 3356 123 42 www.xy433 Proxy[8] 7557 ....

[Article contains additional citation context not shown here]

Z. Wang, "Cachemesh: A Distributed Cache System for World Wide Web", Web Cache Workshop, 1997.

....communication to avoid the storage of equal objects and the distribution of information about stored projects for the allocation process. Previous research on cooperative proxies can be found in the area of hierarchical [27] and hashing approaches [13] 29] adaptive caching [17] CacheMesh [28], WebWave [9] and the straightforward approach with a central coordinator [18] 26] Additionally research for distributed systems cover areas for consistency between multiple proxies [8] the theoretical description of the underlying processes [14] 29] and attempts to build the system on the idea ....

Z. Wang, Cachemesh: A Distributed Cache System for World Wide Web, Web Cache Workshop, 1997.

.... to find cached web pages, due to the fact that their location is pre defined and the search algorithm requires no further overhead, but their major drawback is inflexibility and poor adaptability [15] 21] Additional work for distributed proxy systems, like Adaptive Web Caching [5] and CacheMesh [7], try to overcome specific performance bottlenecks. For example, in Adaptive Web Caching through dynamically created proxy groups combined with data multi casting, while CacheMesh computes the routing protocol based on exchanged routing information. Both approaches can still be considered ....

Zhen Wang, Jon Crowcroft, Cachemesh : A Distributed Cache System for World Wide Web, NLANR Web cache workshop, June 1997

....server Send the beginnings of video segments in the window of focus Meta data key frames Segmented video Window of focus Initiate prefetching User Hierarchical structure of the video Figure 5: The video prefetching scheme. 4.1. 4 Cooperative Caching Cooperative caching [17] [27], 29] is widely studied and deployed in many content delivery systems, starting from Harvest [5] and Squid [24] A number of proxies cooperate and share the cache with one another to improve the cache hit rate. When a proxy receives a request to an object that is missing from its cache, it checks ....

Z. Wang and J. Crowcroft, "Cachemesh: A Distributed Cache System for World Wide Web," Proceedings of the WCW'97, Boulder, Colorado, June 1997.

....directly promote sharing of cached documents. However, the mechanisms used in summary caching for sharing directory information can also be used in informing the client about the availability of documents in the reference point cache of a server. System of Caching Proxies: Harvest [13] Cachemesh [18], Cooperating Web Caches [19] and Adaptive Web Caching [20] are all proposed caching schemes that involve a hierarchy of caches that cooperate. However, they all follow the current model of proxy caching, that is each client goes thorugh exactly one proxy and the proxies fetch and deliver all ....

Z Wang, "Cachemesh: A distributed cache system for world wide web," in Web Cache Workshop, 1997.

....the authors applies the idea to inter ISP cache sharing. The paper propose that each proxy should keep index of only the objects cached in its vicinity in the network. Thus, the overhead for object location is only going to be incurred when it pays o to nd the object. 6. 3 Cachemesh Cachemesh [27] is a distributed cache system that builds upon the work of Harvest cache. Like the Harvest cache, a Cachemesh is comprised of several co operative cache servers. However, this approach achieves the cooperating caching in a little di erent way than the Harvest approach. Central to the Cachemesh ....

Zheng Wang. Cachemesh: A distributed cache system for world wide web. In 2nd NLANR Web Caching Workshop, http://www.belllabs. com/user/zhwang/papers/cache.html, June 1997.

....of web services: web caching service and distributed web service. Caching and replication have been known for long to be the solution to applications performance degradation. As such they were proposed extensively to be applied to the Web (and to FTP) as soon as they become fairly popular [4][5][6] 7] 8] However neither solution has been widely adopted. The main reason has been the fact that both solutions depend very much on network characteristics, require a well thought placement and organization of servers and a coordinated installation and execution, thereby it takes a long time to ....

Wang, Z., Crowcroft, J., "Cachemesh: A Distributed Cache System For World Wide Web," NLANR Web Cache Wk., Boulder, Jun. 1997.

....share meta information indicating the location of Web documents in Web caches. Cache cooperation needs a careful selection of the cooperating caches and in some cases a whole new application layer routing infrastructure with intermediate caches in the path from the local ISP to the origin server [87] [67] Additionally, requests not hit in the local ISP need to travel to other ISP caches, which may be overloaded or connected through congested links. If many documents could be prefetched into local caches, most of the requests could be satisfied locally and no intermediate caches or ....

Z. Wang and J. Crowcroft, "Cachemesh: A Distributed Cache System For World Wide Web", In 2nd International WWW Caching Workshop, Boulder, Colorado, USA, June 1997.

....latency for the request to the origin server, as discussed above. 38 To avoid waiting for a set of responses before passing the request on, some peer groups share content or location information with each other. Cache Digests [RW] WebHints [ISYO] content lists [MNR] or location information [WCr] is shared among the peer caches. This information is used to identify the best next hop for the request, directing each request either to the cache that is most likely to have the document or the one that is closest to the origin server. In the latter case, the intent is that the request will ....

Wang, Z., and Crowcroft, J., "Cachemesh: A Distributed Cache System For World Wide Web," in Proceedings 2 nd NLANR Web Caching Workshop, Boulder, CO, June 1997.

....share meta information indicating the location of Web documents in Web caches. Cache cooperation needs a careful selection of the cooperating caches and in some cases a whole new applicationlayer routing infrastructure with intermediate caches in the path from the local ISP to the origin server [30] [32] Additionally, requests not hit in the local ISP need to travel to other ISP caches, which may be overloaded or connected through congested links. If many documents could be prefetched into local caches, most of the requests could be satisfied locally and no intermediate caches or ....

Z. Wang, "Cachemesh: A Distributed Cache System For World Wide Web", , UCL, June 1997.

....of Web documents in Web caches. Adaptive Web caching [31] proposes a multicast based adaptive caching infrastructure. Requests travel towards the origin server through a series of caches that are automatically selected using a content based routing protocol. A similar approach is used by cachemesh [29] and cache routing [17] with routing tables in every cache or at the network level to route requests not satisfied in the cache. Cache cooperation needs a careful selection of the cooperating caches, and in some cases a whole new applicationlayer routing infrastructure with intermediate caches in ....

Z. Wang, "Cachemesh: A Distributed Cache System For World Wide Web", , UCL, June 1997.

....cache in the path to the origin server. Adaptive Web caching [10] proposes a multicast based adaptive caching infrastructure. Requests travel towards the origin server through a series of caches that are automatically selected using a routing based protocol. A similar approach is used by cachemesh [21] and cache routing [8] with routing tables in every cache or at the network level to route requests not satisfied in the cache. In this paper we do not consider how to best route documents through a series of caches in the path to the origin server. Instead, we present the degree of cooperation ....

Z. Wang and J. Crowcroft, "Cachemesh: A Distributed Cache System For World Wide Web", In 2nd International WWW Caching Workshop, Boulder, Colorado, USA, June 1997.

....returns the code 304 , which stands for document unmodified. When a document is requested through a caching hierarchy some additional delays are introduced: 1) Resolution delay, which is the time to check if the document is present in an ISP (ICP queries [27] hashing function [21] routing [25]) In order to keep this delay low a caching hierarchy should not have more than three levels [7] 2) TCP delay, which is due to the slow start phase of the different TCP connections between every cache level [18] The slow start is more relevant when the completion time of the document is ....

Z. Wang, "Cachemesh: A Distributed Cache System For World Wide Web", , UCL, June 1997.

.... However, going through a caching hierarchy to obtain a document has several additional delays that a multicast distribution has not: ffl Resolution delay: This delay accounts for the time to check if the document is kept by any cache at that level (ICP queries [26] hashing function [21] routing [24]. ffl TCP delay: This delay is due to the slow start phase of the different TCP connections between every cache level [17] The slow start is more relevant when the completion time of the document is small. The effect of this delay gets very reduced when persistent TCP connections are allowed ....

Z. Wang, "Cachemesh: A Distributed Cache System For World Wide Web", , UCL, June 1997.

....intercepted and serviced by the network caches, thus reducing server load. In the case where N=1000, we see a decrease in server load of 54 . Caches shared between sites are clearly beneficial. Ultimately, we would like to compare intra network caching to other systems of shared caches (such as [7, 26, 31]) that locate their caches at network endpoints. However, it is not 0 200 400 600 800 1000 0 200 400 600 800 1000 Server Load (requests second) Client Load (requests second) Effect of Caching on Server Load Client site caching only Plus Active Router Caching Figure 1: Server Load yet clear ....

Z. Wang and J. Crowcroft. Cachemesh: A distributed cache system for world wide web. NLANR Web Caching Workshop, June 1997.

.... However, going through a caching hierarchy to obtain a document has several additional delays that a multicast distribution has not: ffl Resolution delay: This delay accounts for the time to check if the document is kept by any cache at that level (ICP queries [26] hashing function [21] routing [24]. ffl TCP delay: This delay is due to the slow start phase of the different TCP connections between every cache level [17] The slow start is more relevant when the completion time of the document is small. The effect of this delay gets very reduced when persistent TCP connections are allowed ....

Z. Wang, "Cachemesh: A Distributed Cache System For World Wide Web", , UCL, June 1997.

....tree size, the trees for small and or lightly loaded servers may be too big, resulting in many useless application level checks. Also, since trees are random, the client server paths through them are likely to be much longer than the regular routing path from client to server. Wang and Crowcroft [9] describe a preliminary plan to put cache routing tables in caches to specify, for each page or server, where to look next if the local cache does not hold the document. A default route for some documents would help keep table size reasonable. Our approach can be considered to be a ....

Zheng Wang and Jon Crowcroft. Cachemesh: A distributed cache system for the World Wide Web. NLANR Web cache workshop http://ircache.nlanr.net/Cache/Workshop97/agenda.html, June 1997.

No context found.

Wang, Z., Crowcroft, J., "Cachemesh: A Distributed Cache System For World Wide Web," NLANR Web Cache Wk., Boulder, Jun. 1997.

No context found.

Z. Wang, "Cachemesh: A Distributed Cache System For World Wide Web", , UCL, June 1997.

No context found.

Z. Wang and J. Crowcroft, "Cachemesh: A distributed cache system for world wide web," presented at NLANL Web Caching Workshop, Boulder CO, 1997.

First 50 documents

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