IP multicasting comes of age

Many corporations regularly need to transmit information to a specified group of recipients. With new IP multicasting protocols and equipment, reliable and bandwidth-efficient IP multicasting can help get the information where it needs to go.

Potential users include financial service firms looking to distribute trading information to brokers, auto companies providing training to dealer service technicians, and any large corporation seeking to distribute software upgrades.

IP multicast is an ideal alternative to IP broadcast and unicast transmission for these applications. Consider a specialized ticker that transmits packets to 100 traders. Unicasting would require the periodic transmission of 100 identical packets, with many packets traversing the same links. Using broadcasting, all network end stations would receive the packets, regardless of the small recipient group. But multicasting sends the same information only once and only to the intended recipients. The resulting bandwidth savings and scalability inherent in multicast provide a major benefit to network operators.

How does it work?

IP multicasting relies on two mechanisms: a group management protocol to establish and maintain multicast groups, and multicast routing protocols to route packets efficiently. The availability of better versions of these protocols is the primary driver behind multicast scalability.

Internet Group Management Protocol (IGMP) manages packet communication between endstations and their local multicast router, letting them join or leave groups. The most recent version, IGMP Version 3, supports source filtering, which ensures that endstations only receive packets from specified multicast sources, rather than all multicast sources.

When group membership is established, multicast packets, identified through a group address in the destination field of the IP header, are forwarded between routers using multicast routing protocols.

Multicast routing protocols construct distribution trees through the network and perform multicast forwarding. Distribution trees define the path that multicast traffic will take through the network to group members. These paths are based on source trees or shared trees.

The simplest method is a source tree, with its root at the source and branches forming a spanning tree throughout the network. Because this type uses the shortest path through the network, it's also called a shortest path tree (SPT). Source trees guarantee minimal latency, but require more network resources because a separate branch is used to reach every group member.

Shared trees use a single common root placed at some chosen point in the network. This shared root is often called a rendezvous point. Multicast sources send their traffic to the rendezvous point, which forwards traffic down the shared tree to all group members. Shared trees make more efficient use of network resources, but don't necessarily use an optimal path, which may increase packet latency.

Multicast routing protocols also come in two flavors: dense mode and sparse mode. Dense mode protocols use a source tree and are ideal for LANs where group members are distributed densely throughout the network. Sparse mode protocols use a shared tree, thus offering a more scalable solution for building distribution trees when group members are distributed sparsely on a WAN.

The most popular multicast routing protocol is Protocol Independent Multicast-Sparse Mode (PIM-SM), which is being deployed widely across IP WANs. PIM-SM uses an explicit "join" model that blocks multicast traffic forwarding unless it is requested; routers must explicitly join a group to receive multicast traffic for that group. This prevents unnecessary flooding of multicast traffic throughout the network, resulting in greater bandwidth efficiency and multicast scalability.

PIM-SM, a shared tree protocol, also allows switchover from the rendezvous point-based tree (RPT) to the SPT model if a performance threshold is violated, offering the best of both worlds. A router will change from the RPT to the SPT if the multicast traffic it receives exceeds a predefined latency threshold, providing a quality-of-service mechanism for multicast.

IP multicast provides an efficient and simple way for enterprise network managers to distribute information and a significant incentive for service providers to deliver affordable public multicast services in the near future. With the advent of new versions of key protocols, IP multicast is ready to roll.

Lindley is product manager for Agilent Technologies' Advanced Networks Division. She can be reached at tara_lindley@agilent.com.