Universal IP multicast delivery

  • Authors:

  • Beichuan Zhang;Wenjie Wang;Sugih Jamin;Daniel Massey;Lixia Zhang

  • Affiliations:

  • Computer Science Department, University of Arizona, Tucson, AZ 85721-0077, USA;EECS Department, University of Michigan, Ann Arbor, MI 48109-2122, USA;EECS Department, University of Michigan, Ann Arbor, MI 48109-2122, USA;Computer Science Department, Colorado State University, Fort Collins, CO 80523-1873, USA;Computer Science Department, UCLA, Los Angeles, CA 90095-1596, USA

  • Venue:
  • Computer Networks: The International Journal of Computer and Telecommunications Networking - Overlay distribution structures and their applications
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

A ubiquitous and efficient multicast data delivery service is essential to the success of large-scale group communication applications. The original IP multicast design is to enhance network routers with multicast capability [S. Deering, D. Cheriton, Multicast routing in datagram internetworks and extended LANs, ACM Transactions on Computer Systems 8(2) (1990) 85-110]. This approach can achieve great transmission efficiency and performance but also poses a critical dependency on universal deployment. A different approach, overlay multicast, moves multicast functionality to end hosts, thereby removing the dependency on router deployment, albeit at the cost of noticeable performance penalty compared to IP multicast. In this paper we present the Universal Multicast (UM) framework, along with a set of mechanisms and protocols, to provide ubiquitous multicast delivery service on the Internet. Our design can fully utilize native IP multicast wherever it is available, and automatically build unicast tunnels to connect IP Multicast ''islands'' to form an overall multicast overlay. The UM design consists of three major components: an overlay multicast protocol (HMTP) for inter-island routing, an intra-island multicast management protocol (HGMP) to glue overlay multicast and native IP multicast together, and a daemon program to implement the functionality at hosts. In addition to performance evaluation through simulations, we have also implemented parts of the UM framework. Our prototype implementation has been used to broadcast several workshops and the ACM SIGCOMM 2004 conference live on the Internet. We present some statistics collected during the live broadcast and describe mechanisms we adopted to support end hosts behind Network Address Translation (NAT) gateways and firewalls.