2. Introduction to Multicast

2.1. Multicast Overview

IP multicast provides an effective method of many-to-many communication. Delivering unicast datagrams is fairly simple. Normally, IP packets are sent from a single source to a single recipient. The source inserts the address of the target host in the IP header destination field of an IP datagram; intermediate routers (if present) simply forward the datagram towards the target in accordance with their respective routing tables.

Sometimes, distribution needs individual IP packets be delivered to multiple destinations (like audio or video streaming broadcasts). Multicast is a method of distributing datagrams sourced from one or more hosts to a set of receivers that may be distributed over different (sub) networks. This makes delivery of multicast datagrams significantly more complex.

Multicast sources can send a single copy of data using a single address for the entire group of recipients. The routers between the source and recipients route the data using the group address route. Multicast packets are delivered to a multicast group. A multicast group specifies a set of recipients who are interested in a particular data stream and is represented by an IP address from a specified range. Data addressed to the IP address is forwarded to the members of the group. A source host sends data to a multicast group by specifying the multicast group address in the datagram’s destination IP address. A source does not have to register in order to send data to a group nor do they need to be a member of the group.

Routers and Layer 3 switches use the Internet Group Management Protocol (IGMP) to manage membership for a multicast session. When a host wants to receive one or more multicast sessions it will send a join message for each multicast group it wants to join. When a host wants to leave a multicast group, it will send a leave message.

To extend multicast to the Internet, the multicast backbone (Mbone) is used. The Mbone is layered on top of portions of the Internet. These portions, or islands, are interconnected using tunnels. The tunnels allow multicast traffic to pass between the multicast-capable portions of the Internet. As more and more routers in the Internet are multicast-capable (and scalable), the unicast and multicast routing table will converge.

The original Mbone was based on Distance Vector Multicast Routing Protocol (DVMRP) and was very limited. The Mbone is, however, converging around the following protocol set:

  1. IGMP
  2. Protocol Independent Multicast (Sparse Mode) (PIM-SM)
  3. Border Gateway Protocol with multi-protocol extensions (MBGP)
  4. Multicast Source Discovery Protocol (MSDP)

2.2. Multicast Models

This section describes the models which Nokia routers support to provide multicast.

2.2.1. Any-Source Multicast (ASM)

Any-Source Multicast (ASM) is the IP multicast service model defined in RFC 1112, Host Extensions for IP Multicasting. An IP datagram is transmitted to a host group, a set of zero or more end-hosts identified by a single IP destination address (224.0.0.0 through 239.255.255.255 for IPv4). End-hosts can join and leave the group any time and there is no restriction on their location or number. This model supports multicast groups with arbitrarily many senders. Any end-host can transmit to a host group even if it is not a member of that group.

To combat the vast complexity and scaling issues that ASM represents, the IETF is developing a service model called Source Specific Multicast (SSM).

2.2.2. Source Specific Multicast (SSM)

The Source Specific Multicast (SSM) service model defines a channel identified by an (S,G) pair, where S is a source address and G is an SSM destination address. In contrast to the ASM model, SSM only provides network-layer support for one-to-many delivery.

The SSM service model attempts to alleviate the following deployment problems that ASM has presented:

  1. Address allocation — SSM defines channels on a per-source basis. For example, the channel (S1,G) is distinct from the channel (S2,G), where S1 and S2 are source addresses, and G is an SSM destination address. This averts the problem of global allocation of SSM destination addresses and makes each source independently responsible for resolving address collisions for the various channels it creates.
  2. Access control — SSM provides an efficient solution to the access control problem. When a receiver subscribes to an (S,G) channel, it receives data sent only by the source S. In contrast, any host can transmit to an ASM host group. At the same time, when a sender picks a channel (S,G) to transmit on, it is automatically ensured that no other sender will be transmitting on the same channel (except in the case of malicious acts such as address spoofing). This makes it harder to spam an SSM channel than an ASM multicast group.
  3. Handling of well-known sources — SSM requires only source-based forwarding trees, eliminating the need for a shared tree infrastructure. In terms of the IGMP, PIM-SM, MSDP, MBGP protocol suite, this implies that neither the RP-based shared tree infrastructure of PIM-SM nor the MSDP protocol is required. Thus, the complexity of the multicast routing infrastructure for SSM is low, making it viable for immediate deployment. MBGP is still required for distribution of multicast reachability information.
  4. Anticipating that point-to-multipoint applications such as Internet TV will be significant in the future, the SSM model is better suited for such applications.

2.2.3. Multicast in IP-VPN Networks

Multicast can be deployed as part of IP-VPN networks. For details on multicast support in IP-VPNs, refer to the 7450 ESS, 7750 SR, and 7950 XRS Layer 3 Services Guide: Internet Enhanced Services and Virtual Private Routed Network Services.