7th International Conference on Performance Evaluation Methodologies and Tools

Research Article

Analysis of the Shortest Queue First service discipline with two classes

  • @INPROCEEDINGS{10.4108/icst.valuetools.2013.254388,
        author={Fabrice GUILLEMIN and Alain SIMONIAN},
        title={Analysis of the Shortest Queue First service discipline with two classes},
        proceedings={7th International Conference on Performance Evaluation Methodologies and Tools},
        publisher={ICST},
        proceedings_a={VALUETOOLS},
        year={2014},
        month={1},
        keywords={quality of service probabilistic models complex analysis},
        doi={10.4108/icst.valuetools.2013.254388}
    }
    
  • Fabrice GUILLEMIN
    Alain SIMONIAN
    Year: 2014
    Analysis of the Shortest Queue First service discipline with two classes
    VALUETOOLS
    ACM
    DOI: 10.4108/icst.valuetools.2013.254388
Fabrice GUILLEMIN1,*, Alain SIMONIAN1
  • 1: Orange Labs Networks
*Contact email: fabrice.guillemin@orange.com

Abstract

To address the problem of buffer bloat causing latency for time sensitive flows in the Internet, we introduce the Shortest Queue First (SQF) algorithm. This service discipline consists of serving the flow with the least number of backlogged bytes in a buffer. Considering a system with two flows and assuming exponentially distributed service times and Poisson arrivals, we propose a method of computing the Laplace transforms of the workloads in the two virtual queues of the system. We notably derive empty queue probabilities and queue asymptotics. As desired, the analytic results show that SQF performs implicit Head of Line priority for the flow with the smallest traffic intensity.