Andrew M. Odlyzko. Data networks are lightly utilized, and will stay that way. Unpublished, AT&T Labs Research, Florham Park, NJ, 1998.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....as the World Wide Web only five years ago) can introduce new network tra#c characteristics and overall tra#c volume still shows exponential growth. More specifically, today there is an almost exclusive demand for IP services in data networks. Data tra#c is expected to outgrow voice tra#c in 2002 [Odl98a]. Many analysts expect that exponential growth in data communication will continue for the next 10 years at least. The characteristics of multimedia tra#c may enforce the introduction of next generation protocols and hardware, while the ability of e#ciently multiplexing tra#c streams still remains ....

....in the context of a DFN like setting is demonstrated in [FJSP99] However, statistical tra#c modeling (an overview can be found e.g. in [Gor97] is an ongoing field of research and still has to prove its practical relevance. 2.7.1. 5 IP Tra#c Models Measurements on the MCI Internet Backbone show [Odl98a, Odl98b, Odl98c] that the connection facilities have light average utilization (10 15 ) and some private corporate networks seem to have even less (close to 5 ) The congestion experienced by the users is caused mainly by the limits of throughput in the exchange points between networks (peer ing points) and by ....

Andrew Odlyzko. Data Networks are lightly utilized, and will stay that Way. Technical report, At&T Labs - Research, 1998.

....capacity measurements, since few carriers provide detailed data. Further, this type of capacity has a tendency to jump suddenly, as bandwidth is usually increased in large steps (such as going from OC3 to OC12, and then OC48, a phenomenon that contributes to the low utilization of data links [Odlyzko1]) Thus there is little regularity in capacity growth figures. On the other hand, we do find astonoshing regularity in traffic growth, which leads us to propose that a form of Moore s Law applies. In the long run, we expect that capacity will grow slightly faster than traffic, as we explain ....

.... if you build it, they will fill it. Our evidence supports this, but with the important qualification that they will not fill it immediately. That certainly has been the experience in local area networks, LANs. The prevalence of lightly utilized long distance corporate links was noted in [Odlyzko1]. That paper also discussed the vBNS (very High Speed Backbone Network) research network, which was extremely lightly loaded. Here we cite another example of a large network with low utilizations and moderate growth rates. Abilene is the network created by the Internet2 consortium of U.S. ....

[Article contains additional citation context not shown here]

A.M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/amoi.

....demand leading to endemic congestion. When I first proposed PMP [23] I shared this view, but based on knowledge of how many networks are operated, felt that one should strive for maximal simplicity even at the expense of maximal efficiency in use of transport capacity. A recent series of studies [9, 13, 25, 26, 27] have led me to question the basic assumptions that underlie the work on differentiated services. Most of the Internet is very lightly utilized, most of the problems are not caused by link or switch congestion (which is what most QoS measures address) and the tragedy of the commons is much ....

Odlyzko, A. M. Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/amoi.

....size and growth of the Internet, as well as about utilization patterns. This data is then used to justify some speculative predictions about the likely evolution of data networks. 1. Introduction This paper presents some of the highlights of the studies of data networks that are documented in [2, 5, 11, 12, 13] and in a few cases updates them. Much more detail about methodologies and results is available in those papers. This paper and those studies consider only highlevel aggregate measurements of the Internet, and do not look at details of protocols, say. There are many studies of the economics of ....

....are many studies of the economics of the Internet. Most of them are listed in [9, 10, 14] However, they are old (by Internet standards) and none of them answer such basic questions as how large the various parts of the Internet are, and how much they cost. A key point of the investigation of [2, 5, 11, 12, 13] was the need to consider not just the public Internet, but the full universe of data networks and their role in the economy. For simplicity, only networks inside the U.S. were considered. Since costs of transmission are much lower in the U.S. than in most other countries, these networks are ....

[Article contains additional citation context not shown here]

A. M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/amoi.

.... if you build it, they will fill it. Our evidence supports this, but with the important qualification that they will not fill it immediately. That certainly has been the experience in local area networks, LANs. The prevalence of lightly utilized long distance corporate links was noted in [Odlyzko1]. That paper also discussed the vBNS research network, which was extremely lightly loaded. Here we cite another example of a large network with low utilizations and moderate growth rates. Abilene is the network created by the Internet2 consortium of U.S. universities [Dunn] Its backbone consists ....

....Many novel applications require high bandwith to be effective. That (together with some additional factors, such as the high growth rate, lumpy capacity, and pricing structure) contributes to the generally much lower utilization of data networks than of the long distance voice network, [Odlyzko1]. Even on more congested links, it often happens that an increase in capacity does not lead to a dramatic increase in traffic. This is illustrated by several examples. Figure 3.1 shows statistics for the traffic from the public Internet to the University of Waterloo over the last 7 years. This ....

A. M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/amoi.

....leading to endemic congestion. When I first proposed PMP [Odlyzko0] I shared this view, but based on knowledge of how many networks are operated, felt that one should strive for maximal simplicity even at the expense of maximal efficiency in use of transport capacity. A recent series of studies [CoffmanO, FishburnO, Odlyzko2, Odlyzko3, Odlyzko4] have led me to question the basic assumptions that underlie the work on differentiated services. Most of the Internet is very lightly utilized, most of the problems are not caused by link or switch congestion (which is what most QoS measures address) and the tragedy of the commons is much less ....

A. M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/¸amoi.

....in 2000) are not substantiated by detailed analysis and appear incorrect, as that transition is occurring about now, it appears. The analysis of [MutooniT] appears to go astray by assuming utilization rates of data and voice networks are about the same. However, as is shown in the companion paper [Odlyzko1] and discussed at greater length in Section 7 below, data networks typically are used much less intensively than voice networks. Thus network capacities do not represent the amount of traffic those networks carry. Finally, there are claims (such as in some of the presentations of Vint Cerf [Cerf] ....

....links. However, the general opinion seems to be that although at one point this was the main use of private lines, today it is a minor factor. We will therefore ignore it. Traffic on private line networks is much harder to estimate than their capacity. The key point of the companion paper [Odlyzko1] is that conventional capacity utilization estimates, such as those of [Leida] are almost an order of magnitude too high. It is impossible to obtain precise estimates, since no measurements are taken on many lines, and even when there are statistics, those are not released. However, both direct ....

[Article contains additional citation context not shown here]

A. M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/¸amoi.

....the efficiency of data transport and making the Internet less expensive and more useful. On the other hand, many popular techniques, such as some Quality of Service measures and ATM, are likely to be of limited usefulness. 1. Introduction An extensive study of data networks is documented in [CoffmanO, FishburnO, Odlyzko2, Odlyzko3, Odlyzko4]. This paper presents only a brief summary of the results of that study, and concentrates on their implications for present and future data networks. Utilization rates of networks have been strangely absent from most papers on the economics of the Internet, such as those in [MacKieM, McKnightB, ....

....links are priced by their maximal capacity. Furthermore, utilization rates are the primary means by which network managers determine quality of transmission. Therefore it seemed important to consider current utilization rates on the Internet and the resulting costs. Such a study was carried out in [Odlyzko2, Odlyzko4]. It uncovered a number of surprising results. For example, corporations in the U.S. spend more to transmit large files over their packet networks than they would if they used modems over the switched voice network. The primary culprit behind this phenomenon is the low utilization of most data ....

[Article contains additional citation context not shown here]

A. M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/¸amoi.

....2 to 6 attempt to partially fill the gap in published information about the economics of the Internet. Fig. 2 is a sketch of the Internet, with the label Internet attached just to the backbones (as the term is often used) As will be shown in Section 2 (based largely on the companion papers [CoffmanO, Odlyzko2]) these backbones are far smaller than the aggregate of corporate private line networks, whether measured in bandwidth or cost (although not necessarily in traffic) See Table 2 for the sizes of data networks in the U.S. It is taken from [CoffmanO] and effective bandwidth, explained in that ....

....is based on the assumption of endemic congestion. However, when we examine the entire Internet, we find that most of it is uncongested. That the LANs are lightly used has been common knowledge. However, it appears to be widely believed that long distance data links are heavily utilized. The paper [Odlyzko2] (see Section 3 for a summary) shows that this belief is incorrect. Even the backbone links are not used all that intensively, and the corporate private line networks are very lightly utilized. There are some key choke points (primarily the public exchange points, the NAPs and MAEs, and the ....

[Article contains additional citation context not shown here]

A. M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/¸amoi.

....demands for differential treatment of packets. Similar demands are coming from the corporate side. Private line networks use the same IP (Internet protocol) technology, are far larger in aggregate than the public Internet [3] and have been providing high QoS largely through low utilization levels [13]. However, with demand for bandwidth rising, corporate network managers are also demanding tools such as prioritization to ensure higher efficiency of network usage. Differential service quality will inevitably force introduction of more complicated pricing schemes than the present one, since it ....

....of service is a complex one. It is possible to have a lightly utilized network that delivers horrible service, but in general the lower the utilization rate, the better the service. Further, many networks, such as corporate Intranets, are already providing QoS largely through low utilization rates [13]. High quality experimental networks such as vBNS also have very low utilizations. These networks are still operated on the best effort basis, with no explicit guarantees (but with sophisticated traffic engineering tools) Congestion episodes are infrequent enough for this to be acceptable. In ....

[Article contains additional citation context not shown here]

A. M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/¸amoi.

....demand leading to endemic congestion. When I first proposed PMP [10] I shared this view, but based on knowledge of how many networks are operated, felt that one should strive for maximal simplicity even at the expense of maximal efficiency in use of transport capacity. A recent series of studies [2, 11, 12, 13] has raised questions about the basic assumptions that underlie the work on QoS, or at least the backbones of the Internet. Most of the Internet is very lightly utilized, most of the problems are not caused by link or switch congestion (which is what QoS measures address) and the tragedy of the ....

A. M. Odlyzko, Data networks are lightly utilized, and will stay that way. Available at hhttp://www.research.att.com/¸amoi.

No context found.

Andrew M. Odlyzko. Data networks are lightly utilized, and will stay that way. Unpublished, AT&T Labs Research, Florham Park, NJ, 1998.

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