Internet Research Needs Better Models: Topologies

Proposals and Projects

• Modeling Topology of Large Internetworks at Georgia Tech.
  "A primary objective of our work is therefore to support the study of large internetworks through scalable, realistic models of internetwork structure and applications."
  The GT-ITM topology generator.

• Internet Topologies and Simulation at the University of Washington.
  "Our goal is measurement, analysis and realistic generation of Internet topology."
  Rocketfuel:
  "We present new Internet mapping techniques that have enabled us to directly measure router-level ISP topologies."

• Topology Modeling Group at Boston University.
  The BRITE topology generator:
  "BRITE supports multiple generation models including models for flat AS, flat Router and hierarchical topologies. Models can be enhanced by assigning links attributes such as bandwidth and delay."

• The Topology Project at the University of Michigan.
  Papers on "several power-law relationships observed on Autonomous Systems' (AS) connectivity degree, degree frequencies, and the neighborhood size".
  The Inet topology generator:
  "Inet, currently at version 3.0, is an Autonomous System (AS) level Internet topology generator."

• The Tiers topology generator.
  The Tiers topology generator is based on a Transit-Stub model.

• Oliver Heckmann, Topologies for ISP Level Network Simulation.
  "On this page we collect realistic network topologies and information how to generate realistic topologies using topology generators. We concentrate on POP and router level topologies (opposite to AS level topologies), where one topology reflects the network of one provider."

• The SSFnet web page has topology models for Tier-1 Internet provider networks with multiple peering points and for Multi-AS topologies from BGP routing tables.

Topology Generators for NS

• Graham Phillips' web page contains some topology generators using the Stanford Graph Base (SGB) topology format, and sgb2ns, which converts topologies from SGB to ns format.

• Georgia Tech Internetwork Topology Models (GT-ITM) generates topologies using the SGB format, and sgb2ns converts topologies from SGB to ns format.

Other

• David Alderson, Lun Li, Walter Willinger, and John C. Doyle, Understanding Internet Topology: Principles, Models, and Validation, IEEE/ACM ToN, December 2005.
  "For the Internet, an improved understanding of its physical infrastructure is possible by viewing the physical connectivity as an annotated graph that delivers raw connectivity and bandwidth to the upper layers in the TCP/IP protocol stack."

• Oliver Heckmann, Michael Piringer, Jens Schmitt, Ralf Steinmetz, On Realistic Network Topologies for Simulation, SIGCOMM Workshop on Models, Methods and Tools for Reproducible Network Research, 2003.
  This paper considers the BRITE, TIERS, and GT-ITM topology generators.
  "To conclude, Tiers was able ... to produce topologies that had the highest similarity to the real world ISP topologies. ... The level of similarity that could be reached is quite high and indicates that hierarchical topology generators are able to produce realistic POP level topologies."

• Ellen Zegura, New Directions and Half-Baked Ideas in Topology Modeling (ppt), IPAM workshop, June, 2002.
  "BGP analysis - what are the modeling needs?"
  "Peer-to-peer/overlay networks - what are the modeling needs?"
  "What topology models are appropriate for wireless/ad-hoc/sensor networks?"
  "Can a focus on the use of models lead to improved ability to evaluate the quality of models?"

• Anagnostakis, Greenwald, and Ryger, On the Sensitivity of Network Simulation to Topology, MASCOTS 2002.
  "We argue, first, that the barbell topology is not representative of the Internet. In particular, we report that a mesurable fraction of packets pass through multiple congestion points. Second, we argue that the distinction between the barbell topology and more complex topologies is relevant."
  A congestion point is defined not in terms of packet drop/mark rates, but in terms of the distribution of queueing delay. The rest of the paper shows that if you use total aggregate goodput as the metric, then, not too surprisingly, odd things happen in simulation scenarios with multiple congestion points.