Multicast Services

Simply stated, the problem solved by multicast is this: allow one host to talk to several others without bothering everyone else. In a world with limitless resources, the solution is not difficultthe source would replicate the data stream into multiple unicast streams destined for all interested parties. In the real world, this obviously is not a scalable approach. On one hand, it would overburden the source, and on the other, it would have a tremendous impact on the network whenever bandwidth-intensive traffic such as video is transmitted. A scalable alternative is to let the network replicate the data as needed. In a "multicast-aware" network, the data stream is replicated along the way, by the network elements, at points where paths to various destinations converge. A set of dedicated protocols ensures that this data is delivered with optimal utilization of the underlying IP infrastructure, while selecting shortest paths with minimal delay between the source and the destinations.

The demand for multicast services is on the rise, driven by the need for collaborative, real-time information sharing. A wide range of multicast-based applications made their way into both enterprise and service provider types of networks:

• Enterprise networks Distribution of financial information (such as stock quotes), distribution of news, videoconferencing, and delivering content to employees (for example, software updates or facilitate distance learning)

• Service provider Content delivery such as video and audio streaming, collaborative applications such as conferencing for enterprise customers or multiplayer gaming and chat for residential customers

Despite the interest it generates, multicast took a long time and a lot of creative thinking to be developed to the level of a reliable, manageable, and high-performance service offering. Multicast is deployed on top of an existent IP unicast infrastructure and conceptually it deals with the same issues of addressing, routing, and forwarding, but sometimes from a radically different perspective. You can find a detailed presentation of the IPv4 multicast operation and its deployment guidelines in Developing IP Multicast Networks, Volume 1 by Beau Williamson.

Multicast came as an afterthought in IPv4 and it had to deal with many of the limitations built in to the protocol up to that point. Some of these intrinsic constraints become more evident as productivity-enhancing applications and customer requirements drive up the number of networks that deploy multicast services. By contrast, IPv6 multicast gained a prominent role from the start and it was developed alongside unicast. This offers the unique opportunity to make it a ubiquitous service from day one. The operation and benefits of IPv6 multicast are discussed in Chapter 6 "Providing IPv6 Multicast Services."

However, no dramatic changes should be expected; both implementations are built on the same principles. IPv6 took advantage of the larger addresses and larger addressing space while following a more practical approach with respect to multicast routing protocols selection based on the lessons learned with the IPv4 multicast services. The following are some of the issues that IPv4 multicast faces:

• Limited availability of globally unique group addresses. This is an expected problem generated by the limited allocation of IPv4 addresses for multicast purposes. IPv6 provides plenty multicast addresses to facilitate the deployment of the service.

• Limited availability of MAC address for layer 2 multicast address mapping. IPv6 is facing a similar problem.

• Rendezvous Point (RP) scalability problems in large multicast domains. IPv6 provides additional mechanisms that facilitate RP to multicast group mapping.

• Complex mechanisms used to contain multicast control and data traffic within multicast domains. Through address scoping, IPv6 offers elegant new ways to mange the multicast traffic.

• RP source registration information synchronization is done through the MSDP protocol, a protocol that was meant to be a temporary solution for this function. No further development was done on the protocol for a couple of years. No mechanism is yet available in IPv6 to perform these functions.

• The adoption of MPLS by most service providers and deployment of MPLS/VPN services are a challenge for multicast. Currently there is no mechanism available to support label switching of multicast. At best, multicast traffic is forwarded without leveraging the MPLS infrastructure. An additional control mechanism is needed in order to isolate the multicast traffic within each VPN. A stop-gap solution called Multicast VPN (MVPN) is currently offered in Cisco IOS. Nevertheless, with MPLS commonly deployed in Service Provider networks and making its way into the large Enterprise ones, a comprehensive solution to the "multicast over MPLS" problem is important not only for IPv4 but equally important for IPv6.

Note

It is important to point out that MVPN isolates the multicast control plane within each multicast VPN routing and forwarding table (VRF). From a forwarding perspective, the multicast traffic is Generic Routing Encapsulation (GRE) encapsulated and IP forwarded. With MVPN, the multicast traffic is not MPLS labeled even though it might be deployed with MPLS based VPNs.

IPv6 multicast continues to evolve and develop. It adapts to practical market requirements such as Triple Play services and collaborative applications. Chapter 6 discusses the IPv6 multicast protocol set. It also covers the deployment options and improvements with respect to IPv4 multicast, demonstrating that IPv6 is well instrumented to deploy and support multicast services.