|
Abstract This paper presents a method, Pathperf, for Internet wide path performance measurements, using widely deployed web servers as vantage servers. The method works on one end of the path, locates a web server at the other end, and then downloads files from this web server to estimate the path. Module 1 collects web server information on the Internet and uses the BGP routing table along with IP geolocation information to map a unicast IP address to a nearby web server. Module 2 leverages the DNS to deliver web server information to the Pathperf clients. The method was implemented in CERNET with three months of data collection locating 3.5 million web servers in 22 656 different autonomous systems (ASes). Planetlab was used to evaluate the proximity between the end hosts and the mapped web servers in terms of latency. Tests show that in 92.5% of the cases, the latency between a host and a mapped web server was less than 16 ms. Then Pathperf was used to estimate the path bandwidth and latency, with the results compared with the Iperf and King methods. The results show that Pathperf has better availability and comparable accuracy.
|
Keywords
network measurements
link performance
web server
unidirectional measurements
|
Issue Date: 15 April 2014
|
|
|
[1] |
Goldoni E, Schivi M. End-to-end available bandwidth estimation tools, an experimental comparison [J]. J Traffic Monitoring and Analysis, 2010: 171-182.
|
[2] |
Jain M, Dovrolis C. Pathload: A measurement tool for end-to-end available bandwidth [C]//Proceedings of Passive and Active Measurements (PAM) Workshop. Fort Collins, USA: Springer, 2002: 101-112
|
[3] |
Strauss J, Katabi D, Kaashoek F. A measurement study of available bandwidth estimation tools [C]//Proceedings of the 3rd ACM SIGCOMM conference on Internet measurement. Karlsruhe, Germany: ACM, 2003: 39-44.
|
[4] |
Hacker T, Athey B, Noble B. The end-to-end performance effects of parallel TCP sockets on a lossy wide-area network [C]//Parallel and Distributed Processing Symposium. Florida: IEEE Press, 2002: 173-182.
|
[5] |
Lucian P, Ghodsi A, Stoica I. HTTP as the narrow waist of the future Internet [C]//Proceedings of the 9th ACM SIGCOMM Workshop on Hot Topics in Networks. Monterey, CA, USA: ACM, 2010: 31-36.
|
[6] |
Netcraft. Web Server Survey [Z/OL]. [2013-11-02] http://news.netcraft.com/archives/category/web-server-survey/.
|
[7] |
Donnet B, Gueye B, Kaafar M A. A survey on network coordinates systems, design, and security[J]. J Communications Surveys & Tutorials, IEEE, 2010, 12(4): 488-503.
|
[8] |
Gummadi K P, Saroiu S, Gribble S D. King: Estimating latency between arbitrary internet end hosts [C]//Proceedings of the 2nd ACM SIGCOMM Workshop on Internet measurement. Pittsburgh, PA: ACM, 2002: 5-18.
|
[9] |
BAO Congxiao, LI Xing, Jiang Jinpeng, et al.Scalable application-specific measurement framework for high performance network video [C]//17th International workshop on Network and Operating Systems Support for Digital Audio & Video. Urbana-Champaign, IL: ACM, 2007: 87-92.
|
[10] |
Ingmar P, Uhlig S, Kaafar M, et al.IP geolocation databases: unreliable[J]. ACM SIGCOMM Computer Communication Review, 2011, 41(2): 53-56.
|
[11] |
LIU Hongqiang, XIONG Yongqiang, BAO Congxiao, et al.WIND: A Scalable and Lightweight Network Topology Service for Peer-to-Peer Applications [C]//Network Operations and Management Symposium. Osaka, Japan: IEEE, 2010: 797-800.
|
[12] |
Ajay T, Cottrell L, Dunigan T. Measuring end-to-end bandwidth with Iperf using Web100 [C]//Proceedings of Passive and Active Measurement Workshop. La Jolla, CA, USA: Springer, 2003: 145-156.
|
|
Viewed |
|
|
|
Full text
|
|
|
|
|
Abstract
|
|
|
|
|
Cited |
|
|
|
|
|
Shared |
|
|
|
|
|
Discussed |
|
|
|
|