Multicast algorithms in service overlay networks

  • Authors:
  • Dario Pompili;Caterina Scoglio;Luca Lopez

  • Affiliations:
  • Rutgers, The State University of New Jersey, Department of Electrical and Computer Engineering, Piscataway, NJ 08854, USA;Kansas State University, Department of Electrical and Computer Engineering, Manhattan, KS 66506, USA;University of Rome La Sapienza, Department of System Engineering, 00184 Rome, Italy

  • Venue:
  • Computer Communications
  • Year:
  • 2008

Quantified Score

Hi-index 0.24

Visualization

Abstract

Overlay routing enhances the reliability and performance of IP networks since it can bypass network congestion and transient outages by forwarding traffic through one or more intermediate overlay nodes. In this paper, two algorithms for multicast applications in service overlay networks are presented. The first algorithm is tailored for source-specific applications such as live video, software and file distribution, replicated database, web site replication, and periodic data delivery; it builds a virtual source-rooted multicast tree to allow one member in the multicast group to send data to the other members. The second algorithm is tailored for group-shared applications such as videoconference, distributed games, file sharing, collaborative groupware, and replicated database; it constructs a virtual shared tree among group members. The objective of both algorithms is to achieve traffic balancing on the overlay network so as to avoid traffic congestion and fluctuation on the underlay network, which cause low performance. To address these problems, the algorithms actively probe the underlay network and compute virtual multicast trees by dynamically selecting the least loaded available paths on the overlay network. This way, network resources are optimally distributed and the number of multicast trees that can be setup is maximized. Both algorithms can offer service differentiation, i.e., provide QoS at application-layer without IP-layer support. The low computational complexity of the proposed algorithms leads to time and resource saving, as shown through extensive simulation experiments.