13.3 IGMP Protocol

Joining a Multicast Group

Fundamental to multicasting is the concept of a process joining a multicast group on a given interface on a host. (We use the term process to mean a program being executed by the operating system.) Membership in a multicast group on a given interface is dynamic ” it changes over time as processes join and leave the group.

We imply here that a process must have a way of joining a multicast group on a given interface. A process can also leave a multicast group that it previously joined. These are required parts of any API on a host that supports multicasting. We use the qualifier "interface" because membership in a group is associated with an interface. A process can join the same group on multiple interfaces.

The release of IP multicasting for Berkeley Unix from Stanford University details these changes for the sockets API. These changes are also provided in Solaris 2.x and documented in the ip (7) manual pages.

Implied here is that a host identifies a group by the group address and the interface. A host must keep a table of all the groups that at least one process belongs to, and a reference count of the number of processes belonging to the group.

IGMP Reports and Queries

IGMP messages are used by multicast routers to keep track of group membership on each of the router's physically attached networks. The following rules apply.

A host sends an IGMP report when the first process joins a group. If multiple processes on a given host join the same group, only one report is sent, the first time a process joins that group. This report is sent out the same interface on which the process joined the group.
A host does not send a report when processes leave a group, even when the last process leaves a group. The host knows that there are no members in a given group, so when it receives the next query ( next step), it won't report the group.
A multicast router sends an IGMP query at regular intervals to see if any hosts still have processes belonging to any groups. The router must send one query out each interface. The group address in the query is 0 since the router expects one response from a host for every group that contains one or more members on that host.
A host responds to an IGMP query by sending one IGMP report for each group that still contains at least one process.

Using these queries and reports, a multicast router keeps a table of which of its interfaces have one or more hosts in a multicast group. When the router receives a multicast datagram to forward, it forwards the datagram (using the corresponding multicast link-layer address) only out the interfaces that still have hosts with processes belonging to that group.

Figure 13.3 shows these two IGMP messages, reports sent by hosts, and queries sent by routers. The router is asking each host to identify each group on that interface.

Figure 13.3. IGMP reports and queries.

We talk about the TTL field later in this section.

Implementation Details

There are many implementation details in this protocol that improve its efficiency. First, when a host sends an initial IGMP report (when the first process joins a group), there's no guarantee that the report is delivered (since IP is used as the delivery service). Another report is sent at a later time. This later time is chosen by the host to be a random value between 0 and 10 seconds.

Next, when a host receives a query from a router it doesn't respond immediately, but schedules the responses for later times. (We use the plural "responses" because the host must send one report for each group that contains one or more members.) Since multiple hosts can be sending a report for the same group, each schedules its response using random delays. Also realize that all the hosts on a physical network receive all the reports from other hosts in the same group, because the destination address of the report in Figure 13.3 is the group's address. This means that, if a host is scheduled to send a report, but receives a copy of the same report from another host, the response can be canceled . This is because a multicast router doesn't care how many hosts belong to the group ” only whether at least one host belongs to the group. Indeed, a multicast router doesn't even care which host belongs to a group. It only needs to know that at least one host belongs to a group on a given interface.

On a single physical network without any multicast routers, the only IGMP traffic is the reports issued by the hosts that support IP multicasting, when the host joins a new group.

Time-to-Live Field

In Figure 13.3 we noted that the TTL field of the reports and queries is set to 1. This refers to the normal TTL field in the IP header. A multicast datagram with an initial TTL of 0 is restricted to the same host. By default, multicast datagrams are sent with a TTL of 1. This restricts the datagram to the same subnet. Higher TTLs can be forwarded by multicast routers.

Recall from Section 6.2 that an ICMP error is never generated in response to a datagram destined to a multicast address. Multicast routers do not generate ICMP "time exceeded" errors when the TTL reaches 0.

Normally user processes aren't concerned with the outgoing TTL. One exception, however, is the Traceroute program (Chapter 8), which is based on setting the TTL field. Since multicasting applications must be able to set the outgoing TTL field, this implies that the programming interface must provide this capability to user processes.

By increasing the TTL an application can perform an expanding ring search for a particular server. The first multicast datagram is sent with a TTL of 1. If no response is received, a TTL of 2 is tried, then 3, and so on. In this way the application locates the closest server, in terms of hops.

The special range of addresses 224.0.0.0 through 224.0.0.255 is intended for applications that never need to multicast further than one hop. A multicast router should never forward a datagram with one of these addresses as the destination, regardless of the TTL.

All-Hosts Group

In Figure 13.3 we also indicated that the router's IGMP query is sent to the destination IP address of 224.0.0.1. This is called the all-hosts group address. It refers to all the multicast-capable hosts and routers on a physical network. Each host automatically joins this multicast group on all multicast-capable interfaces, when the interface is initialized . Membership in this group is never reported .