Mah, B. An empirical model of http network tra#c. In IEEE Infocom (1997).  Home/Search   Document Not in Database   Summary   Related Articles   Check

....2, 8] examined trac data to characterize the network, protocols or the user behavior. After the birth of the Web, lots of e ort has been devoted to study caching and content delivery architecture, analyzing traces at the application level (typically log les of web or proxy servers) 9 11] In [12], the authors use large traces collected at the university campus at Berkeley to characterize the HTTP protocol. In [13] the authors start from data collected from a large Olympic server in 1996 to understand TCP behavior, like loss recovery eciency and ACK compression. In [14] authors analyzed ....

Mah, B.: An Empirical Model of HTTP Network Trac. IEEE INFOCOM '97 (1997)

....for minimal delay or minimal drop rates in forwarding. The VLL is sourced and sinked by a RealVideo[1] clientserver pair as shown. We use MGEN DREC[2] to generate a Poisson distributed UDP BE background tra#c. thttp[8] generates a HTTP TCP BE background tra#c based on an empirical HTTP tra#c model[21]. The idea is that it is easier to observe the relatively di#erentiated forwarding services in a moderately or heavily flooded network. Further, it is more interesting to attempt to distinguish e#ects on the QoS due to attacks from those due to random fluctuations in the background tra#c. The ....

Bruce A. Mah. An empirical model of HTTP network tra#c. In INFOCOM (2), pages 592-- 600, 1997.

....system regardless the hardware or the operating system. Within this context, it is important to understand how HTTP tra#c behaviors in order to understand and make improvements in the Internet. One way to do this is developing and using HTTP tra#c models. Many works have been made in this area [2, 3, 4, 5, 6], using di#erent approaches in the development. The majority proposes models that describe a common web client behavior [2, 5, 6, 7] with di#erent levels of details. A small number of works [3, 4] focuses on the behavior of a group or aggregation of web clients, which has as main advantage the ....

....in order to understand and make improvements in the Internet. One way to do this is developing and using HTTP tra#c models. Many works have been made in this area [2, 3, 4, 5, 6] using di#erent approaches in the development. The majority proposes models that describe a common web client behavior [2, 5, 6, 7], with di#erent levels of details. A small number of works [3, 4] focuses on the behavior of a group or aggregation of web clients, which has as main advantage the simplicity. These works do not present precise methods to control the network load generated by their models. In many cases this is a ....

[Article contains additional citation context not shown here]

B. Mah, "An empirical model of http network tra#c," in Proc. INFOCOM'97, Apr. 1997.

....pertaining to the use of the transport protocol are discussed in Section 13. Web tra c is closely tied to the content of Web pages, which varies as new Web page design tools and styles, types of content and content encoding schemes are introduced [81] Trace studies of HTTP 1. 0 tra c such as [89, 119], have shown that most request sizes are smaller than 500 bytes, and therefore t in a typical size TCP segment (about 500 bytes) On the other hand, the mean size of a reply (carrying one component of a page) is typically between 10,000 and 20,000 bytes, and the median ranges between 1,000 ....

.... of a page) is typically between 10,000 and 20,000 bytes, and the median ranges between 1,000 and 2,000 bytes [131] This relatively early study found that most Web pages contain fewer than 5 in lined les, have an average size smaller than 32KB, and 90 of them are smaller than about 200KB [119]. In a summary of Web studies [159] an average HTML le size of about 5KB, with a median of 2KB, and an average image size of 14KB are listed. These gures are probably increasing as the network infrastructure improves and users are able to download larger les. In a recent measurement study of ....

[Article contains additional citation context not shown here]

Mah B., An Empirical Model of HTTP Network Trac, in Proceedings of INFOCOM, April 1997.

....majority of trac in a Web client server exchange is from the server sending requested objects back to the client. Thus, the critical step, in terms of evaluating AQM techniques, in developing a realistic Web trac model is to properly re ect the ow characteristics between server and clients. Mah [17] develops an empirical model of HTTP 1.0 trac by analyzing packet traces on a university LAN. He reports the mean object reply size between 8 10 KB, the median reply sizes around 1.5 2.0 KB and the maximum reply size over 1 MB. Mah claims that Web reply size distributions are heavy tailed and can ....

....reply size distribution body can be modeled using a log normal distribution with = 9:357 and = 1:318. Guo and Matta [12] use a bounded Pareto function with = 1:2 to generate Web replies. As described in Section 4. 1, the Web trac generator developed for our investigation combines the model in [17] with results from [2] 5] and [12] 3 SHRED This section presents background issues with short lived ows in Section 3.1 and details on the SHRED architecture in Section 3.2. 3.1 Short lived Flow Issues For both congestion and ow control, TCP uses a congestion window (cwnd) to limit a ....

[Article contains additional citation context not shown here]

B. A. Mah. An Empirical Model of HTTP Network Trac. In Proceedings of INFOCOM, pages 592-600, Apr. 1997.

....values, giving a wide variation in round trip time delays between pairs of clients and servers. Each of the web client nodes models a number of web browsers simultaneously, resulting in several thousand web browers sharing the bottleneck link. The web browsing models are as described by Mah in [24]. This topology matches closely that used in a similar experiment described in [25] The purpose of the experiment is to measure the average end to end delay experienced by web browser clients, to compare the performance of the two queuing disciplines. A representative set of results from our ....

B. A. Mah, \An empirical model of http network trac," in Proceedings of IEEE INFOCOMM, pp. 592-600, 1997.

....Gaussian distribution. The time between the retrieval of two successive Web pages from the same client, namely the user think time, is modeled through a Pareto distribution [9] The number of embedded objects per page request including the base HTML page is also obtained from a Pareto distribution [9, 22]. The inter arrival time of hit requests, that is, the time between retrieval of two successive hit requests from the servers, is modeled by a heavy tailed function distributed as a Weibull. The distribution of the le sizes requested to a Web server is a hybrid function, where the body is modeled ....

B.A. Mah, \An empirical model of HTTP network trac", Proc. of IEEE Int. Conf. on Computer Communication, Kobe, Japan, April 1997.

....this packet represents a user click. The problem is to determine the value of T click . The value should be large enough, so that requests for parts of the same web page is not counted as user clicks, and small enough to separate di erent user clicks. Similar problems have been addressed by Mah [11] and by Crovella and Bestavros [6] When investigating packet traces in order to determine the number of les per web page, Mah uses the threshold value 1 second to separate connections that belongs to di erent web pages. The main reason for the choice of this value was that users will generally ....

....to track all documents referenced by unmodi ed Netscape Naviga tor clients. The third approach of gathering data, and the method used here, is to analyze packet traces taken from a subnet carrying HTTP trac. This method was used by Stevens [15] to analyze the trac arriving at a server, and by Mah [11] to model the client side of the HTTP trac. A further step is taken by Anja Feldman [7] when extracting full HTTP level as well as TCP level traces via packet monitoring. VI. Conclusions and future work We have presented an empirical model for web client traf c. The model is based on user ....

B. A. Mah, \An empirical Model of HTTP network trac," in INFOCOM `97 Conference Proceedings, pp. 592-600, Kobe, Japan april 7-11, 1997.

No context found.

Mah, B. An empirical model of http network tra#c. In IEEE Infocom (1997).

No context found.

B. Mah. An empirical model of http network tra#c. In Proceedings of INFOCOM '97, Kobe, Japan, April 1997.

No context found.

B. Mah, "An empirical model of HTTP network tra#c," in Proceedings of the IEEE Infocom. Kobe, Japan: IEEE, Apr. 1997, pp. 592--600. [Online]. Available: http: //www.ca.sandia.gov/bmah/Papers/Http-Infocom.ps

No context found.

Bruce A. Mah. An Empirical Model of HTTP Network Tra#c. In INFOCOM (2), pages 592--600, 1997. (pp 41, 42, 43)

No context found.

Mah B., An Empirical Model of HTTP Network Trac. Proceedings of INFOCOM 1997, Kobe, Japan, April 1997.

No context found.

Bruce A. Mah. An Empirical Model of HTTP Network Tra#c. In Proceeding of Infocom 1997, pages 592--600, Kobe, Japan, April 1997.

No context found.

B. Mah. An empirical model of http network trac. In Proc. Infocom'97, April 1997. 35

No context found.

B. Mah, An Empirical Model of HTTP Network Trac, Proceedings of INFOCOM '97, Kobe, Japan, April 1997.

No context found.

Mah, B. A. An Empirical Model of HTTP Network Trac. In IEEE INFOCOM (1997).

No context found.

B. Mah, "An empirical model of HTTP network tra#c," in Proceedings of INFOCOM'97, vol. 2, pp. 592--600, IEEE, April 1997.

No context found.

B. Mah. An Empirical Model of HTTP Network Trac. In Proceedings of IEEE Infocom '97, 1997.

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