Vern Paxson. End-to-end routing behavior in the Internet. IEEE/ACM Transactions on Networking, 5(5):601--615, 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....h. Further, R n is not stable in the Internet routing maps change over time in response to routing updates. The former irregularity has significant impact over which links we can directly measure. In contrast, measurements suggest that for our purposes we need not model the latter instability [12]; we verify this assumption in Section III C. B. Technique We can directly estimate the delay on link l = hx; yi as follows. Let s 2 N denote our measurement source, and t x and t y denote the timestamp difference returned from ICMP Timestamp Request packets sent back toback from s to nodes x ....

....path, as we will discuss in Section III D) The robustness of our technique therefore depends on the probability that all paths to internal nodes involved in measurement are stable and not just the end to end path, making our technique more sensitive to path instability. While it is known [12] that around 91 of end to end paths are likely to be stable over periods of hours, this result does not necessarily account for paths towards network internal nodes. Although we are not aware 0 0.2 0.4 0.6 0.8 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 packet pair error (ms) ....

Vern Paxson, "End-to-end routing behavior in the Internet," in Proceedings of ACM SIGCOMM, August 1996, pp. 25--38.

....performance measurement are emerging. Today, there are a wealth of work dedicated to the realm of network performance measurement and analysis. The most related include NIMI [3] IPMA project [4] NAE workshop [5] NAI [6] Surveyor [7] San qi Li s SMAQ tool [8] and Vern Paxson s thesis work [9]. In the analyzing stage, traffic models play a significant role. Accurate models enhance our understanding of the complicated network characteristics and behaviors and allow us to study the effect of various model parameters on network performance through simulation [2] There are many tools to ....

Vern Paxson, End-to-End Routing Behavior in the Internet, In Proceedings of ACM SIGCOMM'96, August 1996.

....performance measurement are emerging. Today, there are a wealth of works dedicated to the realm of network performance measurement and analysis. The most related include NIMI [1] IPMA project [2] NAE workshop [3] NAI [4] Surveyor [5] San qi Li s SMAQ tool [6] and Vern Paxson s thesis work [7, 8,9], and etc. National Internet Measurement Infrastructure (NIMI) is a general architecture intended for facilitating scalable Internet measurement infrastructure. The goal of NIMI is to construct a ubiquitous deployed platform, in which a group of measurement platforms cooperatively measure the ....

. Vern Paxson, End-to-End Routing Behavior in the Internet, In Proceedings of ACM SIGCOMM'96, August 1996

....and published periodically as an AS core poster [10] Construction of IP level Internet maps was also done by Cheswick and Burch [11] who made their traceroutes publicly available in 1999 2001. Analysis. Internet path properties were originally studied by Vern Paxson in his Ph.D. thesis [12]. This and later work by Paxson and colleagues focus considerable attention on handling noise in data, with careful exposition of sources and consequences of errors in various classifications, including with routing pathologies. Their study of stationarity of Internet path properties [13] ....

Vern Paxson, "End-to-End Routing Behavior in the Internet," in IEEE/ACM Transactions on Networking, Oct 1997, vol. 5, pp. 601--615.

....surprising that so many paths have remained exactly the same. We have not examined the causes of path instability in detail, but besides routing anomalies, there is always a certain amount of churn in routing, and technologies like load balancing can cause apparent instability [11] 12] 13] [14]. The percentage of stable paths, if stated in terms of complete traces, is of course higher; for example, 70 of complete a root traces were stable. Table II provides statistics on the prefix coverage of stable paths. The BGP table time is the time of the RouteViews snapshot used for converting ....

....and published periodically as an AS core poster [23] Construction of IP level Internet maps was also done by Cheswick and Burch [24] who made their traceroutes publicly available in 1999 2001. Analysis. Internet path properties were originally studied by Vern Paxson in his Ph.D. thesis [14]. This and later work by Paxson and colleagues focus considerable attention on handling noise in data and upon the sources and consequences of errors in analysis. They also discuss routing pathologies. Their study of stationarity of Internet path properties [12] concluded that many IP paths ....

Vern Paxson, "End-to-End Routing Behavior in the Internet," in IEEE/ACM Transactions on Networking, Oct 1997, vol. 5, pp. 601--615.

....Fluttering refers to rapidly varying routing [P97] More than one router name for a hop indicates that the path has changed during the trace time, which usually lasts tens of seconds. Fluttering may be a side effect of load balancing. However, it might introduce problems described in [P97] and [V97]. In our study we found that fluttering is very frequent in the Internet. Approximately 25 (3011 traceroutes) of all the traces contains at least one hop with two different routers. However, the distribution of fluttering paths is not equal. Nearly half (1501 traceroutes) of the fluttering paths ....

Vern Paxson, "End-to-end routing behavior in Internet," IEEE/ACM transactions on Networking 5, page 601-615 19

....Fluttering refers to rapidly varying routing [P97] More than one router name for a hop indicates that the path has changed during the trace time, which usually lasts tens of seconds. Fluttering may be a side effect of load balancing. However, it might introduce problems described in [P97] and [V97]. In our study we found that fluttering is very frequent in the Internet. Approximately 25 (3011 traceroutes) of all the traces contains at least one hop with two different routers. However, the distribution of fluttering paths is not equal. Nearly half (1501 traceroutes) of the fluttering paths ....

Vern Paxson, "End-to-end routing behavior in Internet," IEEE/ACM transactions on Networking 5, page 601-615.

....one RealPlayer configured to use UDP and the other RealPlayer configured to use TCP. While the testbed network setup did not guarantee that two streams with same source and destination address always traveled the same network path, the relatively persistent characteristics of Internet routing [22] suggest they shared the same route in most cases. Also, any routing changes made during streaming applied to both TCP and UDP streams. The hub facilitated capturing network layer performance since packets destined to either PC were broadcasted to both PCs. We ran tcpdump , a well known ....

Vern Paxson, "End-to-End Routing Behavior in the Internet," IEEE/ACM Transactions on Networking, vol. 5, no. 5, pp. 601-- 615, 1997.

....demonstrated anomalies associated with BGP behavior. Labovitz et al. 32, 33] analyzed statistical occurrences of several unexpected scenarios and tried to explain their origins. They found a surprisingly large percentage of routing information that was useless and even pathological . Paxson [43] uses experimental data to study various properties of the BGP produced routes, such as symmetry, persistence and prevalence. He found that Internet paths are heavily dominated by a single prevalent route, but lifetimes of such routes can vary from seconds to days. It is not clear to what extent ....

Vern Paxson. End-to-end routing behavior in the internet. In Proceedings of ACM SIGCOMM '96 Conference, Stanford University, USA, August 1996.

....and number of hops to each host we count. We are aware of a number of web sites that report round trip times and packet loss to a variety of destinations, for example the Internet Weather Report [8] As far as we can tell, only few projects keep track of the distance (in hops) to hosts (Paxson [9] is an exception) All try to measure the time to reach prede ned hosts, yielding results that are not statistically signi cant for the Internet as a whole. 2.2 Bandwidth measurements Bandwidth measurement is useful as a scienti c tool, as an engineering and maintenance tool, and for protocols ....

....Hosts B and C showed round trip times of 2,000 and 500 ms respectively. Most interestingly, host A shows a very high (20 40 seconds) round trip time for the rst 40 or so packets, then drops to about 400ms. These high numbers and strange behaviors show the need to be alert for routing transients [9] and other phenomena that might a ect our measurements. We elaborate on this in the next section. 3.3 Aberrant Internet Behavior In the course of our measurements, we have made a number of observation of atypical behavior of Internet hosts. All of these could be due to mis implementation but ....

Vern Paxson. End-to-end routing behavior in the internet. Computer Communication Review, 26(4), October 1996. (Proceedings of SIGCOMM '96).

....loss will ultimately lead to a high RTT. The router throughput (e.g. number of packets per second that can be routed) may cause a bottleneck. Sometimes routing algorithms may be a headache, especially in international routes. An end to end routing behavior in the Internet is discussed in Paxson [12]. The fifth candidate for bottleneck is the transmission protocol used to download a Web page. TCP is used for HTTP and FTP in downloading Web pages. TCP has some properties that are not suitable for HTTP. TCP needs three way handshaking for connection establishment and fourway handshaking for ....

Vern Paxson, "End-to-end routing behavior in the Internet," In Proceedings of ACM SIGCOMM '96, Palo Alto, CA, August 1996.

....by NLANR were truncated prematurely, for example, the routing tables from Dec. 1999 are mostly less than half the expected size. We exclude all such truncated tables from our study. Second, we found instances in our data set where an AS (or link) disappears for a period of time. The authors of [25], 26] reported frequent losses of connectivity reachability on the Internet due partly to routing instabilities caused by BGP implementation bugs, hardware glitches, and human errors. These outages can last anywhere from several minutes, to several hours, to longer than a week [27] Outages are ....

Vern Paxson, "End-to-end routing behavior in the internet," in Proceedings of ACM SIGCOMM '96, Aug. 1996, pp. 25--38.

No context found.

Vern Paxson. End-to-end routing behavior in the Internet. IEEE/ACM Transactions on Networking, 5(5):601--615, 1997.

No context found.

Vern Paxson. End-to-end routing behavior in the internet. In SIGCOMM Symposium on Communications Architectures and Protocols, Stanford, California, August 1996.

No context found.

Vern Paxson. End-to-end routing behavior in the internet. Stanford, California, August 1996.

No context found.

Vern Paxson, "End-to-end routing behavior in the Internet," IEEE/ACM Transactions on Networking, vol. 5, no. 5, pp. 601--615, 1997.

No context found.

Vern Paxson. End-to-End Routing Behavior in the Internet. ######## ############ ## #### #######, pages 601-615, December 1998.

No context found.

Vern Paxson. End-to-end routing behavior in the internet. In ACM SIGCOMM Computer Communications Review, volume 26, pages 25--38, October 1996.

No context found.

Vern Paxson. End-to-end routing behavior in the Internet. IEEE/ACM Transactions on Networking, 5(5):601--615, 1997.

No context found.

Vern Paxson. End-to-End Routing Behavior in the Internet. IEEE/ACM Transactions on Networking (TON), 5(5):601--615, 1997.

No context found.

Vern Paxson. End-to-end routing behavior in the Internet. IEEE/ACM Transactions on Networking, 5(5):601-- 615, October 1997.

No context found.

Vern Paxson. End-to-end routing behavior in the Internet. In Proceedings of the ACM SIGCOMM Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications, volume 26,4 of ACM SIGCOMM Computer Communication Review, pages 25-38, New York, August 1996. ACM Press.

No context found.

V. Paxon, "End-to-end routing behavior in the Internet," IEEE/ACM Transactions on Networking, vol. 5, no. 5, 1997.

No context found.

Vern Paxson, "End-to-end routing behavior in the Internet," IEEE/ACM Trans. Networking, vol. 5, no. 5, pp. 601 -- 615, Oct. 1997.

No context found.

Vern Paxson. End-to-end routing behavior in the internet. In Proc. of the ACM SIGCOMM Conference. ACM Press, August 1996.

First 50 documents  Next 50

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