SSFNet Gallery of Network Models
gallery of baseline networks

Standard baseline NMS challenge topology

An internetwork of 20 AS clouds; each AS contains over 500 client and server hosts; clients can initiate hundreds of thousands of sessions; traffic workload is a 50-50 mix of TCP and UDP; heavy-tailed traffic workloads are available.

Internet provider topologies and routing

Large topology network models for analysis of routing dynamics and interactions between BGP and OSPF.

Heavy-tailed traffic workloads

High variability, application-level TCP and HTTP traffic workload models generating long-range dependence in IP packet rate correlations.

Simulation performance analysis

Canonical families of models and guidelines for high-detail simulation performance analysis.

Model validation

Extensive validation testsuites for simulated protocols.

Standard baseline NMS challenge topology
A set of "Lego block" network modules and traffic workloads to build any internetwork of 20 or more Autonomous System clouds with a few configuration changes.

Each AS contains 508 client and server hosts and 29 routers.

The traffic workload can be chosen as fixed-size file transfers or as the 50-50 mix of Pareto-distributed TCP and UDP sessions.

Contact: David Nicol, Dartmouth.

Go to download page.

Internet provider topologies and routing

Tier-1 Internet provider networks with multiple peering points
A new generation of SSFNet models employs the approximate backbone topologies of the largest Internet providers (as published by them). The geographical coordinates of the POPs, Data Centers and Exchanges are used to compute the approximate long-distance link delays.

The models support fully configured routers that run BGP4 (external and internal) and OSPFv2. Two or more geographically overlapping backbones can be configured to peer at the specified peering locations.

These models can be further elaborated by attachment of customer networks at distinct POPs, either for the studies of routing, or to study the traffic, or both.

Author: Michael Liljenstam, Dartmouth with Andy Ogielski, Renesys.

Go to download page.

Multi-AS topologies from BGP routing tables
Prototype generator of multi-AS router-level topologies approximating the customer-provider or peering relationships of Autonomous Systems extracted from the BGP routing tables.

Includes a set of subgraphs of AS graphs obtained from BGP routing tables, stored as adjacency lists; and a Perl script for the generation of the DML files that can be directly used in SSFNet simulations.

Author: BJ Premore, Dartmouth.

Go to download page.

Heavy-tailed traffic workloads
Heavy-tailed TCP sessions
A parsimonious model designed to analyze aggregated traffic of many TCP file transfers, such that each connection has independent packet losses due to competition with other TCP traffic.

The traffic workload models are client-server sessions, generating the configurable ON/OFF renewal processes. Either small fixed-size files, or heavy-tailed distributions of file sizes with a configurable variance of inter-session periods are available.

Author: MIchael Liljenstam, Dartmouth and Andy Ogielski, Renesys.

Go to download page.

Empirical WWW traffic workload
Measurement-based models of a web client and a web server, with object size distributions and ON/OFF renewal process session dynamics following a model in the Appendix of the paper Dynamics of IP traffic: a study of the role of variability and the impact of control by Anja Feldman, Anna Gilbert, Polly Huang and Walter Willinger, in Proc. of SIGCOMM'99.

Source code available in the SSFNet distribution, in the directory src/SSF/OS/WWW/.

Use the annotated example in the DML file src/SSF/OS/WWW/test/httpClientServer.dml in order to employ this workload generation in other topologies.

Simulation performance analysis
Methodologies for performance comparison
Systematic methodology for performance comparison of different simulators, using the classical "dumbbell" topologies and multiple large-size TCP data transfers.

Author: David Nicol, Dartmouth.

Go to download page.

scalability in parallel simulation
The effects of bandwidth, delay, and workload on simulation scalability on multiprocessor machines can be systematically studied using parametric families of models, such as rings or stars of subnetworks.

Source code available in the SSFNet distribution, in the directory examples/parallel.

Model validation
SSF BGP
A suite of SSF BGP validation tests for compliance with RFCs - from simple to fairly complex topologies.

Author: BJ Premore, Dartmouth.

Go to BGP validation pages.

Source code for all tests is in the SSFNet distribution, in the directory src/SSF/OS/BGP4/test.
Documentation is in the directory src/SSF/OS/BGP4/doc.

SSF OSPFv2
A suite of SSF OSPFv2 validation tests for compliance with RFCs; including tests from the InterOperability Lab at the Univ. of New Hampshire.

Author: Dirk Jacob, Univ. Saarbruecken.

Go to OSPFv2 validation pages (local mirror).
(Dirk's website is here).

Source code for all tests is in the SSFNet distribution, in the directory src/SSF/OS/OSPFv2/test.
Documentation is in the directory src/SSF/OS/OSPFv2/doc.

SSF TCP
A suite of SSF TCP validation tests; each test page contains the test description, source code, and analysis of plots of tcpdump data and TCP state variables for compliance with RFCs.

Author: Hongbo Liu, Rutgers; and Andy Ogielski, Renesys.

Go to TCP validation pages.

Also see another set of tests is in the SSFNet distribution, in the directory src/SSF/OS/TCP/test.


Last updated July 6, 2002.
Copyright 2002 SSF Research Network.