I. Wakeman, D. Lewis, and J. Crowcroft, "Traffic Analysis of Trans-Atlantic Traffic," Proceedings of INET ' 92, Kyoto, Japan, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....links. These studies have focussed on the dynamics of packet arrivals on the link [FL91, LTWW94, PF95, WTSW95] the characteristics of packet flows [JR86, He90, CBP95] or on traffic patterns over particularly singular links, such as the trans Atlantic link connecting the U.S. and the U.K. [CW91, WLC92]. For link studies of local area networks [JR86, FL91, LTWW94, WTSW95] heterogeneity presents less of a problem than for those of wide area networks, because the latter encompass a much broader range of traffic sources and path characteristics than the former. Some wide area link studies attempt ....

I. Wakeman, D. Lewis, and J. Crowcroft, "Traffic Analysis of Trans-Atlantic Traffic," Proceedings of INET ' 92, Kyoto, Japan, 1992.

....networks have been in use since the early 1970 s, until recently we have known virtually nothing about the characteristics of the individual connections of different protocols. In the last few years a number of papers have appeared giving statistical summaries of traffic on a per protocol basis [C aceres89, Heimlich90, CW91, EHS92, WLC92], an important first step. The next step in understanding wide area traffic is to form models for simulating and predicting traffic. One such model, tcplib [DJ91, DJCME92] is now available. tcplib is an empirical model of wide area traffic: it models the distribution of the random variables ....

.... 27Nov91 24 hours DEC (DEC 3) Mon 15:02 02Dec91 24 hours coNCert (NC) Wed 09:04 04Dec91 24 hours UK US (UK) Wed 05:00 21Aug91 17 hours Table 3: Summary of Additional Datasets traces from DEC WRL; Wayne Sung provided traces of traffic to from the coNCert network in North Carolina; and the authors of [WLC92] provided their traces of the UK US academic network. The first four traces all originate from stub sites, while the latter two represent inter network traffic (though the authors of [WLC92] characterize the UK side of the UK US traffic as similar to a large stub site since it comprises only a ....

[Article contains additional citation context not shown here]

I. Wakeman, D. Lewis, and J. Crowcroft, "Traffic Analysis of Trans-AtlanticTraffic", Proceedings of INET'92, Kyoto, Japan, 1992.

....backbone [11] and evaluated specific routing approaches for use on the backbone [9] Feldmeier [10] studied the estimated performance of a gateway routing table cache. Caceres et al. 3] and Danzig et al. 8] profiled the characteristics of individual application conversations. Wakeman et al. [15] and Asaba et al. 1] present analyses of trans atlantic and transpacific traffic, respectively. This research is supported by a grant of the National Science Foundation (NCR 9119473) and a joint study agreement with the International Business Machines, Inc. Our characterization is based on ....

I. Wakeman, D. Lewis, and J. Crowcroft. Traffic analysis of trans-atlantic traffic. In Inet '92, June 1992.

....that different layers exhibit workload profiles that are different enough to warrant attention when choosing input parameters or building traffic generators for simulation studies on network and protocol performance. The Internet hierarchy is not limited to domestic U.S. infrastructure. Wakeman [63] and Asaba et al. 64] have published studies of traffic characterization across international infrastructure, specifically trans Atlantic and trans pacific traffic, respectively. These efforts are prerequisite, but only preliminary, to more in depth study of geographic flow characterization. ....

....the duration of the sampling interval. However, network traffic is typically non stationary, and so the effect of spreading the sampled packets over a longer interval is not clear. Previous studies indicate that packet arrivals are not independently distributed through time [28] 5] 6] [63] [64] 24] and so it is likely that 100 packets from a 10 second interval will not provide the same estimate as 100 packets spaced over a 1 minute interval, even when differing utilization may render the number of packets in the two parent populations equal. In order to experiment with the effect ....

I. Wakeman, D. Lewis, and J. Crowcroft, "Traffic analysis of trans-Atlantic traffic," in Proceedings of Inet '92, pp. 417--430, June 1992.

....along with X.500, this ISO protocol s use is waning. mud refers to a multi user network game. Its growth is fitful. We simply note that such games are not a new phenomenon: a study of network traffic across the UK US Academic Network link in August, 1991, attributed 11 of all packets to games [WLC92]. Finally, as mentioned above, other traffic aggregates all the remaining connections whose protocols we did not individually analyze. In LBL 7, for example, we observed about 2,750 distinct TCP responder ports in the other connections. We only attempted to identify the most popular. In ....

I. Wakeman, D. Lewis, and J. Crowcroft, "Traffic Analysis of Trans-Atlantic Traffic", Proceedings of INET'92, Kyoto, Japan, 1992.

....from 143 connections in LBL 5 to 888 in LBL 7. mud refers to a multi user network game. Its growth is fitful, and we simply note that such games are not a new phenomenon: a study of network traffic across the UK US Academic Network link in August, 1991, attributed 11 of all packets to games [WLC92]. Finally, as mentioned above, other traffic aggregates all the remaining connections whose protocols we did not individually analyze. In LBL 7, for example, we observed about 2,750 distinct TCP responder ports in the other connections. We only attempted to identify the most popular. In ....

I. Wakeman, D. Lewis, and J. Crowcroft, "Traffic Analysis of Trans-Atlantic Traffic", Proceedings of INET'92, Kyoto, Japan, 1992.

....gateways. Overall we find that the analytic models provide good descriptions, and generally model the various distributions as well as empirical models. 1 Introduction In the last few years a number of papers have appeared giving statistical summaries of wide area traffic on a per protocol basis [C89, H90, CW91, WLC92, DHS93], an important first step to characterizing WAN traffic. The next step in understanding wide area traffic is to form models for simulating and predicting traffic. One such model, Tcplib [DJ91, DJCME92] is now available. Tcplib is an empirical model of wide area traffic: it models the distribution ....

....of traf 2 The statistics missing for the LBL 2 dataset are due to abnormal termination of the tracing program; this termination, however, did not imply any extra ordinary loss of packets during the 30 day study period. fic to from the coNCert network in North Carolina; and the authors of [WLC92] provided their traces of the UK US academic network. The first four traces all originate from stub (endpoint) sites, while the latter two represent inter network traffic (though the authors of [WLC92] characterize the UK side of the UK US traffic as similar to a large stub site since it ....

[Article contains additional citation context not shown here]

I. Wakeman, D. Lewis, and J. Crowcroft, "Traffic Analysis of Trans-Atlantic Traffic", Proceedings of INET'92, Kyoto, Japan, 1992.

....named as NASA or UK academic organisations, packet types should be specified by the services they provide interactive data, bulk data, video data, audio data, specific sites, protocol suites (DECnet, IP etc. and specific protocols. Our initial class structure definition has the the usage see [5] following hierarchical ordering when we are allocating the shares of the bandwidth: 1. Agency 2. Protocol Suite 3. Protocol 4. Service This partitioning first allows the bandwidth to be divided up by organisation, allowing arbitrary levels in this level. The next level is a suggestion that ....

Ian Wakeman, Dave Lewis & Jon Crowcroft, "Traffic Analysis of trans-Atlantic traffic," Computer Communications 16(June 1993), 376,388.

....as NASA or UK academic organisations, packet types should be specified by the services they provide interactive data, bulk data, video data, audio data, specific sites, protocol suites (DECnet, IP etc. and specific protocols. T1 link split into two equal channels. For details on the usage see [5] Our initial class structure definition has the following hierarchical ordering when we are allocating the shares of the bandwidth: 1. Agency 2. Protocol Suite 3. Protocol 4. Service This partitioning first allows the bandwidth to be divided up by organisation, allowing arbitrary levels in ....

Ian Wakeman, Dave Lewis & Jon Crowcroft, "Traffic Analysis of trans-Atlantic traffic," Computer Communications16 (June 1993), 376,388.

.... should ensure that latencies through the network due to queuing delays are low, whereas if large packets were used IP packets have a bi modal distribution from the size of a TCP acknowledgment in whichever data link encapsulation is required to the maximum transmission unit over the network [11] the queuing latencies would be large. However, the value of these so called advantages is debatable. It is extremely unlikely that the distribution of the packets will scale according to the size of packet. Computers generally manipulate data in large chunks such as a frame of video or a page ....

....of University College London (UCL) The particular choices made in the design of the classes and the classifier are described in Section A.2, the design of the Streams module 1 Currently split into an E1 and a T1 link, with the T1 link split into two equal channels. For details on the usage see [11] and its virtual interface are described in Section A.3 and performance measurements reported in Section A.4. We conclude in Section A.5 with ruminations on the feasibility of this design and the limitations of the design from the Streams mechanism and the pointers to future work. A.2 Link ....

Ian Wakeman, Dave Lewis & Jon Crowcroft, "Traffic Analysis of trans-Atlantic traffic, " Computer Communications 16 (June 1993), 376,388.

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