Running ModelGraft to Evaluate Internet-scale ICN - Université Pierre et Marie Curie Accéder directement au contenu
Communication Dans Un Congrès Année : 2016

Running ModelGraft to Evaluate Internet-scale ICN

Résumé

The analysis of Internet-scale Information-centric networks, and of cache networks in general, poses scalability issues like CPU and memory requirements, which can not be easily targeted by neither state-of-the-art analytical models nor well designed event-driven simulators. This demo focuses on showcasing performance of our new hybrid methodology, named ModelGraft, which we release as a simulation engine of the open-source ccnSim simulator: being able to seamlessly use a classic event-driven or the novel hybrid engine dramatically improves the flexibility and scalability of current simulative and analytical tools. In particular, ModelGraft combines elements and intuitions of stochastic analysis into a MonteCarlo simulative approach, offering a reduction of over two orders of magnitude in both CPU time and memory occupancy, with respect to the purely event-driven version of ccnSim, notably one of the most scalable simulators for Information-centric networks. This demo consists in gamifying the aforementioned comparison: we represent ModelGraft vs event-driven simulation as two athletes running a 100-meter competition using sprite-based animations. Differences between the two approaches in terms of CPU time, memory occupancy, and results accuracy, are highlighted in the score-board.
Fichier non déposé

Dates et versions

hal-01383260 , version 1 (18-10-2016)

Identifiants

  • HAL Id : hal-01383260 , version 1

Citer

Michele Tortelli, D. Rossi, E. Leonardi. Running ModelGraft to Evaluate Internet-scale ICN. ACM ICN, Demo session, Sep 2016, Kyoto, Japan. pp.213-214. ⟨hal-01383260⟩
140 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More