On Active Estimation of End-to-End Capacity: A Comparison (Research Paper Sample)

On Active Estimation of End-to-End Capacity: A Comparison
ABSTRACT
The end-to-end capacity, defined as the maximal transmission rate of the weakest link on the entire path between two end hosts, can be useful in various applications. Popular examples include network tomography for tracking and visualizing Internet topology, capacity planning support and optimized route selection, to name just a few. The purpose of this study is to uniformly classify, experimentally evaluate and compare three end-to-end capacity estimation tools, namely pathwave, PBProbe and pingpair. The controlled testbed built to evaluate the tools consists of a source, a transit and a destination subnetwork, whereas the end-to-end capacity of the path to be measured was adjusted to 100 Mb/s. The performance of each tool has been investigated in terms of estimation error, time and traffic injected into the network to deliver an estimate. The results show that pathwave produces the most satisfactory estimates in terms of achieved estimation accuracy, even in presence of intense cross-traffic scenarios within only two seconds. Particularly, pathwave produces for all testbed experiments an average estimation error of 4.13%. The average estimation errors of PBProbe and pingpair for the same experiments, in contrast, are 12.69% and 30.69%, respectively. Furthermore, it is observed that pathwave and PBProbe are also able to produce acceptable results in presence of interrupt coalescence (IC) condition in the network interface card (NIC), whereas pingpair strongly underestimates the end-to-end capacity of the path in such real-world conditions.
1 INTRODUCTION
Knowledge of bandwidth-related metrics is of great interest for both network operators and end-users as it plays a significant role in efficient network management and operation. Some representative examples where knowledge of bandwidth-related metrics can be useful are network error detection and diagnosis, validation of service level agreements, detection of Denial of Service attacks/anomaly detection, network tomography for tracking and visualizing Internet topology, detection of congested or underutilized links and capacity planning support, admission control policies at massively-accessed content servers, optimized congestion control for reliable transport protocols (e.g. for TCP) and optimized network route selection.
In principle, in bandwidth estimation research there are three popular metrics: capacity, available bandwidth and throughput. The capacity states the maximum number of bits per time unit a network link can theoretically transfer. The available bandwidth of a network link is defined as the average residual capacity of that link in a given time period. Finally, throughput, in turn, can be categorized in achievable throughput or bulk transfer capacity (BTC). Achievable throughput is the maximum number of bits per time unit that a link can provide to an application, given the current utilization, the transport protocol and operating system used, and the end-host performance capability and load. In contrast, BTC is a TCP specific metric exhibiting the maximal throughput attainable by a single TCP connection. Each of these metrics can be estimated either on the entire path between two end-hosts (i.e. at the end-to-end scope) or hop-by-hop. For a formal and detailed definition of these metrics, the interested reader is referred to the respective literature ADDIN CSL_CITATION {"citationItems":[{"id":"ITEM-1","itemData":{"DOI":"10.1109/MNET.2003.1248658","ISBN":"2075946302","ISSN":"0890-8044","abstract":"In a packet network, the terms bandwidth and throughput often characterize the amount of data that the network can transfer per unit of time. Bandwidth estimation is of interest to users wishing to optimize end-to-end transport performance, overlay network routing, and peer-to-peer file distribution. Techniques for accurate bandwidth estimation are also important for traffic engineering and capacity planning support. Existing bandwidth estimation tools measure one or more of three related metrics: capacity, available bandwidth, and bulk transfer capacity. Currently available bandwidth estimation tools employ a variety of strategies to measure these metrics. In this survey we review the recent bandwidth estimation literature focusing on underlying techniques and methodologies as well as open source bandwidth measurement tools.","author":[{"dropping-particle":"","family":"Prosad","given":"R.","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Davrolis","given":"C.","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Murray","given":"Margaret","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Claffy","given":"K.C. C","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Prasad","given":"Ravi","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Dovrolis","given":"Constantinos","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Murray","given":"Margaret","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Claffy","given":"K.C. C","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Association","given":"Cooperative","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Analysis","given":"Data","non-dropping-particle":"","parse-names":false,"suffix":""}],"container-title":"IEEE Network","id":"ITEM-1","issue":"6","issued":{"date-parts":[["2003"]]},"page":"27-35","title":"Bandwidth estimation: metrics, measurement techniques, and tools","type":"article-journal","volume":"17"},"uris":["http://www.mendeley.com/documents/?uuid=e378d644-90da-4382-a3f0-b5158c3d69f3"]},{"id":"ITEM-2","itemData":{"DOI":"10.1007/s11277-015-2459-2","ISSN":"0929-6212","abstract":"The ability to locate bottleneck link along Internet paths is of great interest to end users as well as network oper a- tors. End users can use the information to estimate performa nce of a network path to a given destination. The knowledge of bottlenecks can provide network operators insight into the causes of congestion and ways of circumventing it. Flows that share a common bottleneck resource can benefit from cooperative congestion control strategies ([16]). Due to the intriguin g nature of the problem and its practical importance, a number of techniques as well as tools have been proposed to measure bandwidth on Internet paths. In this paper we survey a large number of publicly available tools and techniques to estima te bandwidth in stub networks. We found that although there are some tools available that can be adapted to meet our requirem ent, these tools do not give consistent performance. We therefor e devise a simple strategy that provides an upper bound to the estimation results. Our scheme ensures that the results of t hese tools do not stray beyond acceptable limits. The methodolog y was validated in actual Internet paths","author":[{"dropping-particle":"","family":"Sairam","given":"Ashok Singh","non-dropping-particle":"","parse-names":false,"suffix":""}],"container-title":"Wireless Personal Communications","id":"ITEM-2","issued":{"date-parts":[["2015"]]},"page":"1425-1476","title":"Survey of Bandwidth Estimation Techniques","type":"article-journal"},"uris":["http://www.mendeley.com/documents/?uuid=717ec87d-6ffc-40ff-ae24-d6056f28fe76"]}],"mendeley":{"formattedCitation":"[1, 2]","plainTextFormattedCitation":"[1, 2]","previouslyFormattedCitation":"[1, 2]"},"properties":{"noteIndex":0},"schema":"https://github.com/citation-style-language/schema/raw/master/csl-citation.json"}[1, 2].
Given the variety of motivations and metrics to be estimated, a plethora of bandwidth estimation tools has been developed in recent years. The extensive study in this research field revealed that there are currently over hundred different tools for estimation of bandwidth-related metrics. Still, several new tools are being published annually. Fig. 1 exemplifies the evolution of some important tools for bandwidth estimation over the last two decades.
Consequently, several performance and comparative studies of various estimation tools have been conducted in this area over the last two decades. Although studies on comparing available bandwidth and throughput measurement tools are manifold ADDIN CSL_CITATION {"citationItems":[{"id":"ITEM-1","itemData":{"DOI":"10.1007/978-3-540-31966-5_24","ISSN":"03029743","abstract":"In this paper we present results of a series of bandwidth estimation experiments conducted on a high-speed testbed at the San Diego Supercomputer Center and on OC-48 and GigE paths in real world networks. We test and com- pare publicly available bandwidth estimation tools: abing, pathchirp, pathload, and Spruce.We also tested Iperf which measures achievable TCP throughput. In the lab we used two different sources of known and reproducible cross-trafficin a fully controlled environment. In real world networks we had a complete knowl- edge of link capacities and had access to SNMP counters for independent cross- traffic verification.We compare the accuracy and other operational characteristics of the tools and analyze factors impacting their performance.","author":[{"dropping-particle":"","family":"Shriram","given":"Alok","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Murray","given":"Margaret","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Hyun","given":"Young","non-dropping-particle":"","parse-names":false,"suffix":""},{"dropping-particle":"","family":"Brownlee","given":"Nevil","non-drop...