R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web proxy caching: the devil is in the details", Proceedings of the Workshop on Internet Server Performance, Madison, Wisconsin, June 1998.  @ NUS  Home/Search   Document Details and Download   Summary   Related Articles   Check

....explicit instrumentation of pages and introduces additional HTTP trac combined with the requirement that browsers accept Javascript. The Web server can also measure the number of connection aborts and resets from the client whichmay suggest the presence of poorly connected, impatient clients [7]. 2.3 Additional Factors for Characterization There are several other factors that could be used in classifying a client. Information in the client request header such as accepted contenttypes, the HTTP protocol version [12] and the client software itself could all be exploited in classifying ....

Ramon Caceres, Fred Douglis, Anja Feldmann, Gideon Glass, and Michael Rabinovich. Web proxy caching: the devil is in the details. In ######## ## ######## ###### ###########, Madison, Wisconsin USA, June 1998.

....management. HTTP headers include caching pragmas that affect whether or not an object may be cached, and if so, for how long. Web caching helps to alleviate load on servers and backbone links, and can also serve to decrease object access latencies. Much research has focused on Web proxy caching [4, 5, 7, 11, 12] and, more recently, on coordinating state among multiple, cooperating proxy caches [13, 30, 33] some of these proposals aim to create global caching structures [27, 34] The results of these studies generally indicate that cache hit rates of 40 50 are achievable, but that hit rate increases ....

....systems, and for the potentials for the use of caching in our environment. We focus on this latter issue in the next section. 6 The Potential Role of Caching in CDNs and P2P Systems Caching in the web is well understood: caches have been shown to absorb bandwidth and reduce access latency [4, 5, 7, 11, 12]. In this section of the paper, we explore caching in the context of the Akamai CDN and Kazaa Cache Object Hit Rate vs. Population 0 20 40 60 80 100 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 Object Hit Rate WWW Ideal WWW Practical Akamai Ideal Akamai Practical ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Workshop on Internet Server Performance, June 1998.

....management. HTTP headers include caching pragmas that affect whether or not an object may be cached, and if so, for how long. Web caching helps to alleviate load on servers and backbone links, and can also serve to decrease object access latencies. Much research has focused on Web proxy caching [4, 5, 7, 11, 12] and, more recently, on coordinating state among multiple, cooperating proxy caches [13, 30, 33] some of these proposals aim to create global caching structures [27, 34] The results of these studies generally indicate that cache hit rates of 40 50 are achievable, but that hit rate increases ....

....systems, and for the potentials for the use of caching in our environment. We focus on this latter issue in the next section. 6 The Potential Role of Caching in CDNs and P2P Systems Caching in the web is well understood: caches have been shown to absorb bandwidth and reduce access latency [4, 5, 7, 11, 12]. In this section of the paper, we explore caching in the context of the Akamai CDN and Kazaa Figure 13. WWW and Akamai object hit rates vs. population: object hit rates as a function of population size. An ideal cache treats all documents as cacheable. A practical cache accounts for HTTP 1.1 ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Workshop on Internet Server Performance, June 1998.

.... varieties of prefetching, and more recently, Content Distribution Networks (CDNs) What has become increasingly clear in recent years is that the traditional target of research, poor eviction and replacement algorithms, is not in fact a serious obstacle to good use of a caching infrastructure [1]. In the current web, many cache eviction events and uncacahable resources are driven by two server application goals: First, providing clients with a recent view of the state of the application (i.e. information that is not too old) Second, providing clients with a consistent view of the ....

R. Caceres, F. Douglis, A. Feldman, G. Glass, and M. Rabinovich, "Web proxy caching: The devil is in the details, " in ACM SIGMETRICS Performance Evaluation Review, Dec. 1998.

....with respect to the request routing table size is also presented. Keywords: autonomy oriented computation, self organization, adaptive proxy server 1. Introduction The Internet is growing exponentially and web caches have been shown to be a feasible way to reduce the overall network tra#c [3, 6, 18] . Web servers store web objects, which are requested by clients spread over the global network. A web cache, or proxy, is usually placed between clients making requests and web servers servicing the requests. It will try to resolve the needed object by its local cache. The Internet tra#c is ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. Performance Evaluation Review, 26(3):11--15, December 1998.

....World Wide Web have turned to a variety of caching techniques to reduce latency and conserve bandwidth [40, 2] These techniques have proven to mask temporary server and network failures and alleviate network hotspots. Many studies have shown the effects on latency of caching content near clients [22, 4]; other studies have shown the benefits of sharing caches among user populations [11] Researchers have sought techniques to share content among regional or national user populations with the goal of reducing access latency for all users. Since the presentation of the Harvest web cache [5] a ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: the devil is in the details, 1998.

....World Wide Web have turned to a variety of caching techniques to reduce latency and conserve bandwidth [40, 2] These techniques have proven to mask temporary server and network failures and alleviate network hotspots. Many studies have shown the e ects on latency of caching content near clients [22, 4]; other studies have shown the bene ts of sharing caches among user populations [11] Researchers have sought techniques to share content among regional or national user populations with the goal of reducing access latency for all users. Since the presentation of the Harvest web cache [5] a ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: the devil is in the details, 1998.

....This has already put great stress on the networks that support all these requests for web objects. One of the solutions to alleviate this problem is to store some frequently referenced web objects in a local machine so that the need to retrieve the same object from its host web site is reduced [3]. The Internet trac is expected to be reduced and the response to user request is expected to be improved. The local machine is commonly known as the proxy cache server. It has a cache of considerable size for storing frequently used web objects. With the increasing demand for information from the ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. Performance Evaluation Review, 26(3):11-15, December 1998.

.... may retrieve a document from the proxy cache if the server is unavailable (due to server crash, congestion, or network partitioning) About 52 of the web traffic is cacheable [5] according to the HTTP [6] cache control mechanisms) resulting in a hit ratio of about 35 (byte hit ratio about 30 ) [7]. Nevertheless, if servers would be more cooperative using the HTTP cache control mechanisms, up to 76 of the documents would be cacheable [5] gaining higher hit ratios. In the domain of Pervasive Computing the temporal locality in the users browsing behavior will result in even more hits than ....

....1 bit (b w) WBMP 1KB WAP phone (Siemens S35i) 5. RELATED RESEARCH Related research to our work actually spans different research areas. On the one hand this is web proxy caching research. This research area has been investigated for quite a long time and many aspects have been deeply examined [3, 4, 5, 7, 13, 15, 16, 17, 18, 19]. The idea of proxy caching goes back to the CERN proxy [15] It was further developed in the Harvest project [4] which also created the concepts of hierarchical caching, the basis of our system scenario. The subsequent open source project Squid [3] has further developed the Harvest ideas and ....

R. Caceres, et al., "Web Proxy Caching: The Devil is in the Details", Proc. of the Workshop on Internet Server Performance, Madison, Wisconsin USA, Jun 1998.

....imposing low overhead on the participating nodes. 1. INTRODUCTION Web caching is a widely deployed technique to reduce the latency observed by web browsers, decrease the aggregate bandwidth consumption of an organization s network, and reduce the load incident on web servers on the Internet [5, 11, 22]. Web caches are often deployed on dedicated machines at the boundary of corporate networks, and at Internet service providers. This paper presents an alternative for the former case, in which client desktop machines themselves cooperate in a peer to peer fashion to provide the functionality of a ....

....object cacheability information that it obtains from requests and responses. In practice, di erent proxy servers decide whether an object is cacheable based on some combination of checks for cache control headers, existence of cookies (in HTTP 1. 0) password protection, cgi bin in the URL, etc [5, 21, 24]. For simulation purposes, we use the available information to approximately deduce cacheability; our main goal is to derive an understanding of Squirrel behavior on diverse workloads, rather than make a case for web caching. Thus we de ne static objects as those accessed with a HTTP GET request ....

[Article contains additional citation context not shown here]

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: the devil is in the details. In Proceedings of the Workshop on Internet Server Performance, Madison, WI, May 1999.

....aid in modeling user behavior. While they achieve a high degree of accuracy with their measurements, their method requires access to privileged net work information. Some studies have used shared web cache logs and domain proxy logs in order to measure various aspects of web performance [6] 3] [4]. An advantage of this method is the ready availability of the logs. There are some ch aw backs, however. First, the measurements are made on a machines which are not co located with the user, and the effects of network characteristics between the user and the point of measurement are not ....

l. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. labinovich. Web Proxy Caching: The Devil is in the Details. In Proceedings of Workshop on Internet Server Performance, Madison, Wisconsin, June 1998.

....used by us to provide guarantees. Section 5 gives the experimental evaluation of our architecture. Section 6 concludes the paper and discusses avenues for future work. 2 Related Work The issue of scalable on demand static content replication and distribution has been addressed in numerous papers [1, 2, 3, 4, 5]. In comparison, related research is relatively scarce when dealing with dynamic web content. However, recently there has been considerable interest in attempting to cache dynamic content to improve the scalability and availability of web based information services [6, 7, 8, 9, 10, 11] The methods ....

R. Caceres, F. Douglis, A. Feldman, and M. Rabinovich, "Web proxy caching: The devil is in the details, " in Proc. of the Workshop on Internet Server Performance, Madison, WI, Jun 1998.

....User Perceived Web Performance, Web Proxy, Web Caching, Content Distribution Networks 1 Introduction user perceived Web performance. There has been extensive work evaluating the macroscopic impact of Web infrastructure for improving overall Web performance, such as caching systems (e.g. [2, 6, 7, 9, 16, 17, 19, 21, 25, 30, 32, 34]) prefetching systems (e.g. 4, 11, 15, 18, 27, 29] and to a limited extent content distribution networks (e.g. 20, 22] Surprisingly, however, there has been relatively little work characterizing the impact on Web performance as perceived by the user of various Web delivery systems. ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Workshop on Internet Server Performance, pages 111--118,

....explicit instrumentation of pages and introduces additional HTTP traffic combined with the requirement that browsers accept Javascript. The Web server can also measure the number of connection aborts and resets from the client which may suggest the presence of poorly connected, impatient clients [7] . 2.3 Additional Factors for Characterization There are several other factors that could be used in classifying a client. Information in the client request header such as accepted content types, the HTTP protocol version [11] and the client software itself could all be exploited in ....

Ramon Caceres, Fred Douglis, Anja Feldmann, Gideon Glass, and Michael Rabinovich. Web proxy caching: the devil is in the details. In Workshop on Internet Server Performance, Madison, Wisconsin USA, June 1998.

....processes, has also been introduced. The benefits of this approach include improved recovery speed and the ability to hide client failures from servers. It can be combined with other application level proxies such as WAP gateways [2] HTTP proxies for both active [15, 20, 11, 19] and passive [8, 18, 3] content, web clipping proxies [16] Request Request Response Response HTML Script Encoder Decoder GUI Engine Client Server Proxy Request Request Response Response HTML Script Encoder Decoder GUI Engine Client Server Proxy Figure 1. Application level proxy. distributed object ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass and M. Rabinovich. Web Proxy Caching: The Devil is in the Details. In Proceedings of

....project at the University of Wisconsin, Madison, is exploring how caching can be applied to dynamic documents [7] Their motivation is that the increasing amount of personalized content makes caching such information difficult and not practical with current proxy designs. Indeed, a recent study [5] of a large ISP trace revealed that over 30 of client HTTP requests contained cookies, which are HTTP header elements typically indicating that a request be personalized. As web servers become more sophisticated and customizable, and as one to one marketing e commerce strategies proliferate the ....

R. Caceres, F. Douglis, A. Feldmann, and M. Rabinovich. Web proxy caching: The devil is in the details. Proceedings of the Workshop on Internet Server Performance, 1998.

.... cache effectiveness have shown that dynamic content accounts for over 50 of all requests [40] A related study of HTTP traces from AT T WorldNet Internet service provider has revealed that 30 of all user requests carry cookies, header elements typically indicating that a request be personalized [2]. The second trend has been a significant increase in on themove access to web content using a variety of mobile resource constrained devices. The growing popularity of web enabled PDAs and cellular phones as well as the explosive growth in Internet startups trying to mobilize web content is a ....

....for the first problem and have the effect of moving content closer to the client. However, these typically do not work well with dynamically generated and personalized content. In the AT T study mentioned earlier, caches can only achieve object hit ratios around 35 and byte hit ratios around 30 [2]. Transcoding is a popular solution to resolve server client mismatches, but is unable to benefit from caching in general, with the result that clients see larger access times. Furthermore, neither technique provides particularly good solutions for coping with user mobility: caches are effective ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Proceedings of ACM SIGMETRICS Internet Server Performance Workshop, June 1998.

....latency for retrieving Web documents. Caching is one means of improving the performance and scalability of the Web. Caching can be applied at several locations: at the client [13] 19] 26] 38] at the server [7] 8] 13] and within the network [1] 2] 5] 9] 10] 13] 14] [18], 21] 35] 36] 38] 41] The latter approach involves caching proxies [1] 18] 41] which are the focus of this paper. Caching proxies have gained widespread popularity on the Internet. By keeping local copies of documents requested by Web clients, and using them to satisfy future ....

....and scalability of the Web. Caching can be applied at several locations: at the client [13] 19] 26] 38] at the server [7] 8] 13] and within the network [1] 2] 5] 9] 10] 13] 14] 18] 21] 35] 36] 38] 41] The latter approach involves caching proxies [1] [18], 41] which are the focus of this paper. Caching proxies have gained widespread popularity on the Internet. By keeping local copies of documents requested by Web clients, and using them to satisfy future requests to the same documents, caching proxies can reduce the amount of traffic flowing ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web Proxy Caching: The Devil is in the Details", ACM Performance Evaluation Review, vol. 26, no. 1, pp. 11-15, December 1998.

....problems. One approach is based on caching copies of Web objects closer to the requesting clients. This may be controlled by servers that push popular Web objects out to other cooperating servers [4, 16] or may be triggered by individual client requests passing through client and proxy caches [5, 7, 26, 27, 28]. Another approach is to use prefetching to reduce response times, by hiding server and network latency [6, 24] A complementary approach is to make the Web server more powerful through the use of a clustered architecture, in which multiple machines function cohesively as a single Web server ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web Proxy Caching: The Devil is in the Details", Performance Evaluation Review, Vol. 26, No. 3, pp. 11-15, December 1998.

....The request is neither GET nor HEAD. Pragma Request: The request includes Pragma: no cache in its header. Authorization (Auth) The request header includes an Authorization header for access control. Older caching proxies would not cache responses to requests that included cookies [4, 7]. Since Squid [1] version 2 will cache such HTTP 1.1 responses, we do not consider requests that include cookies to be uncacheable. In addition, the response may have uncacheable properties: Cache Control: The HTTP 1.1 response header includes a Cache control header that prohibits caching. ....

....reductions for each approach is shown in Table 4. The Web proxy alone only eliminates 14 of the bytes. That this is so small does not appear to be solely an artifact of the short time scale of the traces, rather that proxy cache hit rates are often cited in terms of documents, instead of bytes [4]. In fact, the Web proxy hit rate by document we observe approaches 25 . Compression yields similar bene ts and operates independently of caching. Redundancy suppression provides still greater bene ts that are not captured by proxy caching alone. Given this result, we believe that our technique ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Proceedings of the Workshop on Internet Server Performance, June 1998.

....of documents, and reasons why documents are not cacheable. Finally, Section 7 summarizes our study and its results. 2 Previous Work Numerous recent studies of web traffic have been performed. These studies include analyses of web access traces from the perspective of browsers [10, 19] proxies [9, 2, 17, 12, 15, 11, 22, 5, 6, 4, 14], and servers [21, 1, 3] The earlier tracing studies were rather limited in request rate, number of requests, and diversity of population. The most recent tracing studies have been larger and generally more diverse. In addition to static analysis, some studies have also used trace driven cache ....

....number of requests, and diversity of population. The most recent tracing studies have been larger and generally more diverse. In addition to static analysis, some studies have also used trace driven cache simulation to characterize the locality and sharing properties of these very large traces [2, 17, 12, 15, 5, 14], and to study the effects of cookies, aborted connections, and persistent connections on the performance of proxy caching [5, 14] In this paper, we expand on these previous research efforts. Our focus is on sharing and cacheability; however we can also compare our current HTTP traffic ....

[Article contains additional citation context not shown here]

Ramon Caceres, Fred Douglis, Anja Feldmann, Gideon Glass, and Michael Rabinovich. Web proxy caching: the devil is in the details, June 1998.

....predict quality of service and performance levels in a network dominated by data traffic a seismic shift in modelling approaches will therefore be required. Pending development of accurate traffic models a more heuristic approach has been applied. A large proportion of web objects are static [4], so caching popular files nearer to the users reduces demand on remote network devices, and minimises the probability of overload at bottlenecks due to bursts in demand for a particular object. WWW caches have therefore been deployed globally in an effort to decrease the load on network and ....

....can only arise from a process, or set of processes, which have memory, so that a stimulus that occurs at one moment may trigger another related event at another later time. Several causes have been proposed for this including heavy tailed distributions for session lengths [3] file lengths [4], and packetization. However, a further cause is required to explain the long timescale dependency we have observed. With the notable exceptions of web spiders and robot assisted browsing, all web requests emanate from a human user. We have shown [5] that individual user traces can exhibit ....

R Caceres, F Douglis, A Feldmann, G Glass and M Rabinovich. Web Proxy Caching: The Devil is in the Detail. 1st Workshop on Internet Server Performance in conjunction with ACM SIGMETRICS. 1998. P111-118

.... access any available document in the Web (i.e. the document set is very large) For Web servers, the requests are restricted to a limited set of documents (i.e. the documents present at the Web server) However, some recent studies have shown that many static documents are never cached [19, 88]. There are three main reasons for these documents not being cached. First, a signi cant fraction of the responses received from the Web servers contain no Last Modified dates, disabling caching of these requests [88] Second, many 43 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 Fraction of ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovinch, \Web Proxy Caching: The Devil is in the Details", Performace Evaluation Review, Vol. 26, No. 1, pp. 11-15, December 1998.

....saw more than 10 . The most widely used CGI script among the sites surveyed was a page counter. In addition to the relatively small overall usage, the ratio of CGI traffic to regular traffic did not grow during the measurement intervals. In 1998, we begin to see a different trend emerge. The [Caceres et al. 98] study collected network traces from the AT T Worldnet ISP for a 12 day period in August of 1997. The authors point out that the conclusion of the [Gribble et al. 97] study was flawed because they did not consider cookies as a reason that certain resources might be uncacheable. They found that ....

....cache. Squid does allow caching of requests that contain cookies, even though it goes against the original Netscape specification. This poses a problem for the conclusion that cookies are a significant factor that prevents caching of resources. Feldmann et al. 99] perform a followup to the 10 [Caceres et al. 98] study, and show that including the requests with cookies, the overall rates of uncacheability was 43 for the ISP trace, and 37 for a trace of the AT T Labs research community from 1997. In [Wills et al. 99b] the authors use active probes to investigate the cacheability of resources from ....

[Article contains additional citation context not shown here]

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Proc. of the Workshop on Internet Server Performance, Jun 1998.

....dynamic, the utility of Web caches has been examined. ASAWF] suggested that proxies may have a limited benefit. Caching and multicast were compared in [RBR] with the conclusion that proxy groups are sufficient unless the docu 39 ments change constantly. Finally, an analysis of proxy caching [CDFGR] found that most user latency comes from the connection setup overhead. Cross domain cache cooperation can be applied to reduce this overhead. 6.2: Anticipation and Aggregation Anticipation has been proposed both to reduce server work and to improve client latency. The Oceans group at Boston ....

Caceres, R., Douglis, F., Feldmann, A., Glass, G., and Rabinovich, M., "Web Proxy Caching: The Devil is in the Details, " presented at Workshop on Internet Server Performance, Madison, WI, June 1998.

.... access any available document in the Web (i.e. the document set is very large) For Web servers, the requests are restricted to a limited set of documents (i.e. the documents present at the Web server) However, some recent studies have shown that many static documents are never cached [29, 30]. There are three main reasons for these documents not being cached. First, a signi cant fraction of the responses received from the Web servers contain no Last Modified dates, disabling caching of these requests [30] Second, many requests in the proxy traces are 0 0.2 0.4 0.6 0.8 1 0 0.2 ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovinch, \Web Proxy Caching: The Devil is in the Details", Performace Evaluation Review, Vol. 26, No. 1, pp. 11-15, December 1998.

....problems. One approach is based on caching copies of Web objects closer to the requesting clients. This may be controlled by servers that push popular Web objects out to other cooperating servers [4, 13] or may be triggered by individual client requests passing through client and proxy caches [5, 7, 20, 21, 22]. Another approach is to use prefetching to reduce response times, by hiding server and network latency [6, 18] A complementary approach is to make the Web server more powerful through the use of a clustered architecture, in which multiple machines function cohesively as a single Web server ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web Proxy Caching: The Devil is in the Details", Performance Evaluation Review, Vol. 26, No. 3, pp. 11-15, December 1998.

....Eindhoven, Postbus 513, 5600 MB Eindhoven Universiteit Utrecht, Postbus 80010, 3508 TA Utrecht # University of Oxford, St. Catherine s College, Manor Road, Oxford OX1 3UJ, UK Universiteit Maastricht, Postbus 616, 6200 MD Maastricht 1 in the network (first proposed by [6] see also [3]) Typically, such an approach is attractive when an organization (like a company or a university) is responsible for (a part of) a network ( 7] This paper deals with the latter subject called proxy placement in [10] given a network with capacitated edges, external demand (request rates for ....

....depicted infrastructure of Fig. 2. E,j contributing overflow edges nearest proxy containing j j (in order of decreasing # # E,j ) 4 A # , B # ) C # ,D # ,E # # permutation of [6] N ] say [10] 28) 6] 23) 2 D # ,E # C [1] 60) 5] 50) 1 E # D [2] 100) 0 none E [4] 0) [3] (0) Then r E, 10] 4 28 = 112, r E, 1] 2 60 = 120, r E, 2] 1 100 = 100 and r E, 4] 0 0=0, so [1] is a candidate to be put in stack at node E. However, there may be a proxy i 0 with even bigger maximal r i 0 ,j . 8 all [j] # the outside world A # A # # # # # # # # # # # # # # # B ....

[Article contains additional citation context not shown here]

R. Caceres, F. Douglis, A. Feldmann, G. Glass and M. Rabinovich, Web proxy caching: the devil is in the details, Performance Evaluation Review 26 (1998), 11 -- 15.

....not cached at proxies. We assumed that all documents with a cgi bin or in the URL string represent dynamic content. These documents account for less than 2 of the HTTP requests recorded in the logs. The nal step in our data reduction process was to identify uncacheable (static) documents [8, 27, 28]. Since the access logs do not provide the required information (e.g. last modi ed dates, set cookie headers, no cache pragmas) we developed an ad hoc strategy to lter the uncacheable documents. We replayed the (partly) reduced access logs through our proxy cache simulator (described later) ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, M. Rabinovich, Web Proxy Caching: The Devil is in the Details, Performance Evaluation Review 26 (1) (1998) 11-15.

....accomodate the request (the problem with this approach is that most browsers do not handle redirection requests properly) A comparison of our scheme Web caching requires the termination of the connection at intermediate points in the network. Since cache hit rates today are of the order of 30 [CAC] the client often encounters additional delay to connect to the server. Our scheme ensures that the connection is terminated only once at a server that can fulfill the request. Our work is strongly motivated by work on open signaling and control plane technologies, both programmable [Laz, Cam, ....

Ramon Caceres, Fred Douglis, Anja Feldmann, Michael Rabinovich, "Web Proxy Caching: The Devil is in the Details," Proceedings of the Workshop on Internet Server Performance, Madison, WI, June 23 1998 to appear.

....query was used as a super query recently, we believe that it will probably be used as a super query for subsequent queries soon if users are doing query refinement. This can be thought as temporal locality. 4. EXPERIMENTS 4. 1 On Excite Query Trace Many web caching studies have used real traces ([3], 9] 16] or synthetic web workloads ( 1] However, these real traces or generated workloads usually do not include CGI script requests or queries. What we really needed was a trace that recorded user queries to a specific database web server. Fortunately we obtained a real query trace of ....

Ramon Caceres, Fred Douglis, Anja Feldmann, Gideon Glass, and Michael Rabinovich. Web Proxy Caching: The Devil Is in the Details. In Workshop on Internet Server Performance, 1998.

.... events such as SYN and FIN packets ffl TCP and or HTTP requests that are not processed by the server ffl Amount of data transferred on aborted responses The value of packet traces has been demonstrated in recent studies on the impact of TCP dynamics on the performance of Web proxies and servers [8, 9]. A complete collection of packet traces of both request and response traffic at a Web server would provide a unique opportunity to gauge how a change to HTTP 1.1 would affect the workload. In addition, the extended Web server logs can allow us to measure the components of delay at the server, and ....

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web proxy caching: The devil is in the details," in Proc. Workshop on Internet Server Performance, June 1998. http://www.cs.wisc.edu/~cao/WISP98.html.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web proxy caching: the devil is in the details", Proceedings of the Workshop on Internet Server Performance, Madison, Wisconsin, June 1998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Workshop on Internet Server Performance, June 1998.

No context found.

R. Caceres, et al. Web Proxy Caching: The Devil is in the Details. ACM SIGMETRICS Performance Evaluation Rev., 26(3):11--15, December 1998. papers/wisp98.ps.gz

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web Proxy caching: The Devil is in the Details", SIGMETRICS Performance Evaluation Review, 261998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web Proxy caching: The Devil is in the Details", SIGMETRICS Performance Evaluation Review, 26, 1998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web Proxy Caching: the Devil is in the Details. In Workshop on Internet Server Performance, Madison, Wisconsin USA., June 1998. (p 32)

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Workshop on Internet Server Performance, June 1998.

No context found.

R. C`aceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Proceedings of the ACM SIGMETRICS Workshop on Internet Server Performance, 1998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Workshop on Internet Server Performance, June 1998.

No context found.

Ramon Caceres, Fred Douglis, Anja Feldman, Gideon Glass, and Micahel Rabinovich. Web proxy caching: The devil is in the details. In ACM SIGMETRICS Performance Evaluation Review, December 1998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web proxy caching: The devil is in the details. In Proceedings of the Workshop on Internet Server Performance, December 1998.

No context found.

Ramon Caceres, Fred Douglis, Anja Feldmann, Gideon Glass, and Michael Rabinovich. Web proxy caching: The Devil is in the details. Performance Evaluation Review, 26(3):11--15, December 1998. Proceedings of the Workshop on Internet Server Performance.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web Proxy Caching: The Devil is in the Details". Performance Evaluation Review, Vol. 26, No. 3, pp. 11-15, December 1998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich. Web Proxy Caching: The Devil is in the Details. In Workshop on Internet Server Performance, June 1998.

No context found.

Ramon Caceres, Fred Douglis, Anja Feldman, Gideon Glass, and Micahel Rabinovich, Web proxy caching: The devil is in the details, ACM SIGMETRICS Performance Evaluation Review, December 1998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web Proxy Caching: The Devil is in the Details," in Proc. Workshop on Internet Server Performance, June 1998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, \Web Proxy Caching: The Devil is in the Details," in Performance Evaluation Review, vol. 26, no. 1, pp. 11-15, December 1998.

No context found.

R. Caceres, F. Douglis, A. Feldmann, G. Glass, and M. Rabinovich, "Web Proxy Caching: The Devil is in the Details", Performance Evaluation Review, Vol. 26, No. 3, pp. 11-15, December 1998.

First 50 documents  Next 50

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

CiteSeer.IST at NUS - Copyright Penn State and NEC. Hosted by the School of Computing, National University of Singapore.