Towards Realistic Million-Node Internet Simulations

Cowie, James
Liu, Hongbo
Liu, Jason
Nicol, David
Ogielski, Andy

Abstract

This paper describes the Scalable Simulation Framework (SSF), a discrete event modeling API designed for the construction and simulation of very large networks. SSF can execute detailed simulations of complex topology networks with a million or more concurrent TCP/IP ows. We describe the overall architecture of SSF, the architecture of the network modeling layers above the basic API structure, and discuss preliminary performance gures for up to 100,000 concurrent TCP fows in a test topology.

Keywords

large scale
discrete event modeling
SSF

Notes

A presentation of SSF, a discrete event API with several implementations. Discusses performance figures with up to 100,000 TCP flows.

It is important since SSF seems to be the largest scale TCP network simulator around. It is parallel, but only currently only up to 14 nodes on a shared memory Sun Sparc. It uses very lightweight threads, to manage network entities. It is a conservative PDES implementation.

It uses a self configuring mechanism to build networks, thus saving difficulty in building huge networks (especially supplying parameters).

Related Papers

Bibtex

 @inproceedings {cowie_nicol.million99,
   author = { Cowie, James and Liu, Hongbo and Liu, Jason and Nicol, David and Ogielski, Andy},
    title = {Towards Realistic Million-Node Internet Simulations},
 booktitle= {Proceedings of the 1999 International Conference on Parallel and Distributed Processing Techniques and Applications},
    month = {Jun.},
     year = {1999}    
}         

Back to Intro By Author By Importance By Keyword By Title By Reference