2nd International ICST Conference on Simulation Tools and Techniques

Research Article

ns-2 Distributed Clients Emulation: Accuracy and Scalability

Download534 downloads
  • @INPROCEEDINGS{10.4108/ICST.SIMUTOOLS2009.5595,
        author={Stein Kristiansen and Thomas Plagemann},
        title={ns-2 Distributed Clients Emulation: Accuracy and Scalability},
        proceedings={2nd International ICST Conference on Simulation Tools and Techniques},
        publisher={ICST},
        proceedings_a={SIMUTOOLS},
        year={2010},
        month={5},
        keywords={ns-2 Performance Evaluation Network Simulation Network Emulation},
        doi={10.4108/ICST.SIMUTOOLS2009.5595}
    }
    
  • Stein Kristiansen
    Thomas Plagemann
    Year: 2010
    ns-2 Distributed Clients Emulation: Accuracy and Scalability
    SIMUTOOLS
    ICST
    DOI: 10.4108/ICST.SIMUTOOLS2009.5595
Stein Kristiansen1,*, Thomas Plagemann1,*
  • 1: University of Oslo, Gaustadallèen 23 N-0371 Oslo, Norway
*Contact email: steikr@ifi.uio.no, plageman@ifi.uio.no

Abstract

ns-2 is a well known network simulator, recently extended with improvements to its emulation fa- cility. Real-time constraints and the boundary be- tween real-world and simulated entities impose scal- ability and accuracy limitations, and distort the sim- ulated network as perceived by the involved real- world applications. This paper presents results from a performance evaluation of the ns-2 emulation fa- cility. Conducting emulation experiments of differ- ing magnitudes, and under varying emulation envi- ronment set-ups, we unveil central types of scala- bility limitations and obtainable accuracy. We find throughput limits using high and low end comput- ers, and a significant throughput decrease when in- creasing the number of involved real-world applica- tions. We furthermore show how end-to-end delay increases both with traffic load and an increasing number of involved real-world applications. More- over, during these conditions, we find that the sys- tem treats these applications increasingly unfair by distributing total throughput unevenly between them, and by imposing different amounts of end-to-end de- lay.