Parallel Network Simulation under Distributed Genesis

  • Authors:
  • Boleslaw K. Szymanski;Yu Liu;Rashim Gupta

  • Affiliations:
  • Department of Computer Science, RPI, Troy, NY;Department of Computer Science, RPI, Troy, NY;Department of Computer Science, RPI, Troy, NY

  • Venue:
  • Proceedings of the seventeenth workshop on Parallel and distributed simulation
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

We describe two major developments in the General Network Simulation Integration System (Genesis): the supportfor BGP protocol in large network simulations and distributionof the simulation memory among Genesis componentsimulations.Genesis uses a high granularity synchronization mechanismbetween parallel simulations simulating parts of a network.This mechanism uses checkpointed simulation stateto iterate over the same time interval until convergence.It also replaces individual packet data for flows crossingthe network partitions with statistical characterization ofsuch flows over the synchronization time interval. We hadachieved significant performance improvement over the sequential simulation for simulations with TCP and UDP traffic. However, this approach can not be used directly to simulatedynamic routing protocols that use underlying networkfor exchanging protocol information, as no packets are exchangedin Genesis between simulated network parts. Wehave developed a new mechanism to exchange and synchronizeBGP routing data among distributed Genesis simulators.The extended Genesis allows simulations of more realisticnetwork scenarios, including routing flows, in additionto TCP or UDP data traffic.Large memory size required by simulation software hindersthe simulation of large-scale networks. Based on ournew support of distributed BGP simulation, we developedan approach to construct and simulate networks on distributedmemory using Genesis simulators in such a way that each participating processor possesses only data relatedto the part of the network it simulates. This solutionsupports simulations of large-scale networks on machineswith modest memory size.