Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?


Example: Configuring Automatic IP Multicast Without Explicit Tunnels

Understanding AMT

Automatic Multicast Tunneling (AMT) facilitates dynamic multicast connectivity between multicast-enabled networks across islands of unicast-only networks. Such connectivity enables service providers, content providers, and their customers to participate in delivering multicast traffic even if they lack end-to-end multicast connectivity.

AMT is supported on MX Series Ethernet Services Routers with Modular Port Concentrators (MPCs) that are running Junos 13.2 or later. AMT is also supported on i-chip based MPCs. AMT supports graceful restart (GR) but does not support graceful Routing Engine switchover (GRES).

AMT dynamically establishes unicast-encapsulated tunnels between well-known multicast-enabled relay points (AMT relays) and network points reachable only through unicast (AMT gateways). Figure 1 shows the Automatic Multicast Tunneling Connectivity.

Figure 1: Automatic Multicast Tunneling ConnectivityAutomatic Multicast Tunneling Connectivity

The AMT protocol provides discovery and handshaking between relays and gateways to establish tunnels dynamically without requiring explicit per-tunnel configuration.

AMT relays are typically routers with native IP multicast connectivity that aggregate a potentially large number of AMT tunnels.

The Junos OS implementation supports the following AMT relay functions:

  • IPv4 multicast traffic and IPv4 encapsulation

  • Well-known sources located on the multicast network

  • Prevention of denial-of-service attacks by quickly discarding multicast packets that are sourced through a gateway.

  • Per-route replication to the full fan-out of all AMT tunnels desired

  • The ability to collect normal interface statistics on AMT tunnels

Multicast sources located behind AMT gateways are not supported.

AMT supports PIM sparse mode. AMT does not support dense mode operation.

AMT Applications

Transit service providers have a challenge in the Internet because many local service providers are not multicast-enabled. The challenge is how to entice content owners to transmit video and other multicast traffic across their backbones. The cost model for the content owners might be prohibitively high if they have to pay for unicast streams for the majority of their subscribers.

Until more local providers are multicast-enabled, there is a transition strategy proposed by the Internet Engineering Task Force (IETF) and implemented in open source software. This strategy is called Automatic IP Multicast Without Explicit Tunnels (AMT). AMT involves setting up relays at peering points in multicast networks that can be reached from gateways installed on hosts connected to unicast networks.

Without AMT, when a user who is connected to a unicast-only network wants to receive multicast content, the content owner can allow the user to join through unicast. However, the content owner incurs an added cost because the owner needs extra bandwidth to support the unicast subscribers.

AMT allows any host to receive multicast. On the client end is an AMT gateway that is a single host. Once the gateway has located an AMT relay, which might be a host but is more typically a router, the gateway periodically sends Internet Group Management Protocol (IGMP) messages over a dynamically created UDP tunnel to the relay. AMT relays and gateways cooperate to transmit multicast traffic sourced within the multicast network to end-user sites. AMT relays receive the traffic natively and unicast-encapsulate it to gateways. This allows anyone on the Internet to create a dynamic tunnel to download multicast data streams.

With AMT, a multicast-enabled service provider can offer multicast services to a content owner. When a customer of the unicast-only local provider wants to receive the content and subscribes using an AMT join, the multicast-enabled transit provider can then efficiently transport the content to the unicast-only local provider, which sends it on to the end user.

AMT is an excellent way for transit service providers (who can get access to the content, but do not have many end users) to provide multicast service to content owners, where it would not otherwise be economically feasible. It is also a useful transition strategy for local service providers who do not yet have multicast support on all downstream equipment.

AMT is also useful for connecting two multicast-enabled service providers that are separated by a unicast-only service provider.

Similarly, AMT can be used by local service providers whose networks are multicast-enabled to tunnel multicast traffic over legacy edge devices such as digital subscriber line access multiplexers (DSLAMs) that have limited multicast capabilities.

Technical details of the implementation of AMT are as follows:

  • A three-way handshake is used to join groups from unicast receivers to prevent spoofing and denial-of-service (DoS) attacks.

  • An AMT relay acting as a replication server joins the multicast group and translates multicast traffic into multiple unicast streams.

  • The discovery mechanism uses anycast, enabling the discovery of the relay that is closest to the gateway in the network topology.

  • An AMT gateway acting as a client is a host that joins the multicast group.

  • Tunnel count limits on relays can limit bandwidth usage and avoid degradation of service.

AMT is described in detail in Internet draft draft-ietf-mboned-auto-multicast-10.txt, Automatic IP Multicast Without Explicit Tunnels (AMT).

AMT Operation

AMT is used to create multicast tunnels dynamically between multicast-enabled networks across islands of unicast-only networks. To do this, several steps occur sequentially.

  1. The AMT relay (typically a router) advertises an anycast address prefix and route into the unicast routing infrastructure.

  2. The AMT gateway (a host) sends AMT relay discovery messages to the nearest AMT relay reachable across the unicast-only infrastructure. To reduce the possibility of replay attacks or dictionary attacks, the relay discovery messages contain a cryptographic nonce. A cryptographic nonce is a random number used only once.

  3. The closest relay in the topology receives the AMT relay discovery message and returns the nonce from the discovery message in an AMT relay advertisement message. This enables the gateway to learn the relay's unique IP address. The AMT relay now has an address to use for all subsequent (S,G), entries it will join.

  4. The AMT gateway sends an AMT request message to the AMT relay's unique IP address to begin the process of joining the (S,G).

  5. The AMT relay sends an AMT membership query back to the gateway.

  6. The AMT gateway receives the AMT query message and sends an AMT membership update message containing the IGMP join messages.

  7. The AMT relay sends a join message toward the source to build a native multicast tree in the native multicast infrastructure.

  8. As packets are received from the source, the AMT relay replicates the packets to all interfaces in the outgoing interface list, including the AMT tunnel. The multicast traffic is then encapsulated in unicast AMT multicast data messages.

  9. To maintain state in the AMT relay, the AMT gateway sends periodic AMT membership updates.

  10. After the tunnel is established, the AMT tunnel state is refreshed with each membership update message sent. The timeout for the refresh messages is 240 seconds.

  11. When the AMT gateway leaves the group, the AMT relay can free resources associated with the tunnel.

Note the following operational details:

  • The AMT relay creates an AMT pseudo interface (tunnel interface). AMT tunnel interfaces are implemented as generic UDP encapsulation (ud) logical interfaces. These logical interfaces have the identifier format ud-fpc/pic/port.unit.

  • All multicast packets (data and control) are encapsulated in unicast packets. UDP encapsulation is used for all AMT control and data packets using the IANA reserved UDP port number (2268) for AMT.

  • The AMT relay maintains a receiver list for each multicast session. The relay maintains the multicast state for each gateway that has joined a particular group or (S,G) pair.

Configuring the AMT Protocol

To configure the AMT protocol, include the amt statement:

You can include this statement at the following hierarchy levels:

  • [edit protocols]

  • [edit logical-systems logical-system-name protocols]

  • [edit routing-instances routing-instance-name protocols]

  • [edit logical-systems logical-system-name routing-instances routing-instance-name protocols]


In the following example, only the [edit protocols] hierarchy is identified.

The minimum configuration to enable AMT is to specify the AMT local address and the AMT anycast prefix.

  1. To enable the MX Series router to create the UDP encapsulation (ud) logical interfaces, include the bandwidth statement and specify the bandwidth in gigabits per second.
  2. Specify the local address by including the local-address statement at the [edit protocols amt relay family inet] hierarchy level.

    The local address is used as the IP source of AMT control messages and the source of AMT data tunnel encapsulation. The local address can be configured on any active interface. Typically, the IP address of the router’s lo0.0 loopback interface is used for configuring the AMT local address in the default routing instance, and the IP address of the router’s lo0.n loopback interface is used for configuring the AMT local address in VPN routing instances.

  3. Specify the AMT anycast address by including the anycast-prefix statement at the [edit protocols amt relay family inet] hierarchy level.

    The AMT anycast prefix is advertised by unicast routing protocols to route AMT discovery messages to the router from nearby AMT gateways. Typically, the router’s lo0.0 interface loopback address is used for configuring the AMT anycast prefix in the default routing instance, and the router’s lo0.n loopback address is used for configuring the AMT anycast prefix in VPN routing instances. However, the anycast address can be either the primary or secondary lo0.0 loopback address.

    Ensure that your unicast routing protocol advertises the AMT anycast prefix in the route advertisements. If the AMT anycast prefix is advertised by BGP, ensure that the local autonomous system (AS) number for the AMT relay router is in the AS path leading to the AMT anycast prefix.

  4. (Optional) Enable AMT accounting.
  5. (Optional) Specify the AMT secret key timeout by including the secret-key-timeout statement at the [edit protocols amt relay] hierarchy level. In the following example, the secret key timeout is configured to be 120 minutes.

    The secret key is used to generate the AMT Message Authentication Code (MAC). Setting the secret key timeout shorter might improve security, but it consumes more CPU resources. The default is 60 minutes.

  6. (Optional) Specify an AMT tunnel device by including the tunnel-devices statement at the [edit protocols amt relay] hierarchy level.
  7. (Optional) Specify an AMT tunnel limit by including the tunnel-limit statement at the [edit protocols amt relay] hierarchy level. In the following example, the AMT tunnel limit is 12.

    The tunnel limit configures the static upper limit to the number of AMT tunnels that can be established. When the limit is reached, new AMT relay discovery messages are ignored.

  8. Trace AMT protocol traffic by specifying options to the traceoptions statement at the [edit protocols amt] hierarchy level. Options applied at the AMT protocol level trace only AMT traffic. In the following example, all AMT packets are logged to the file amt-log.

    For AMT operation, configure the PIM rendezvous point address as the primary loopback address of the AMT relay.

Configuring Default IGMP Parameters for AMT Interfaces

You can optionally configure default IGMP parameters for all AMT tunnel interfaces. Although, typically you do not need to change the values. To configure default IGMP attributes of all AMT relay tunnels, include the amt statement:

You can include this statement at the following hierarchy levels:

  • [edit protocols igmp]

  • [edit logical-systems logical-system-name protocols igmp]

  • [edit routing-instances routing-instance-name protocols igmp]

  • [edit logical-systems logical-system-name routing-instances routing-instance-name protocols igmp]

The IGMP statements included at the [edit protocols igmp amt relay defaults] hierarchy level have the same syntax and purpose as IGMP statements included at the [edit protocols igmp] or [edit protocols igmp interface interface-name] hierarchy levels. These statements are as follows:

  • You can collect IGMP join and leave event statistics. To enable the collection of IGMP join and leave event statistics for all AMT interfaces, include the accounting statement:

  • After enabling IGMP accounting, you must configure the router to filter the recorded information to a file or display it to a terminal. You can archive the events file.

  • To disable the collection of IGMP join and leave event statistics for all AMT interfaces, include the no-accounting statement:

  • You can filter unwanted IGMP reports at the interface level. To filter unwanted IGMP reports, define a policy to match only IGMP group addresses (for IGMPv2) by using the policy's route-filter statement to match the group address. Define the policy to match IGMP (S,G) addresses (for IGMPv3) by using the policy's route-filter statement to match the group address and the policy's source-address-filter statement to match the source address. In the following example, the amt_reject policy is created to match both the group and source addresses.

  • To apply the IGMP report filtering on the interface where you prefer not to receive specific group or (S,G) reports, include the group-policy statement. The following example applies the amt_reject policy to all AMT interfaces.

  • You can change the IGMP query interval for all AMT interfaces to reduce or increase the number of host query messages sent. In AMT, host query messages are sent in response to membership request messages from the gateway. The query interval configured on the relay must be compatible with the membership request timer configured on the gateway. To modify this interval, include the query-interval statement. The following example sets the host query interval to 250 seconds.

    The IGMP querier router periodically sends general host-query messages. These messages solicit group membership information and are sent to the all-systems multicast group address,

  • You can change the IGMP query response interval. The query response interval multiplied by the robust count is the maximum amount of time that can elapse between the sending of a host query message by the querier router and the receipt of a response from a host. Varying this interval allows you to adjust the number of IGMP messages on the AMT interfaces. To modify this interval, include the query-response-interval statement. The following example configures the query response interval to 20 seconds.

  • You can change the IGMP robust count. The robust count is used to adjust for the expected packet loss on the AMT interfaces. Increasing the robust count allows for more packet loss but increases the leave latency of the subnetwork. To modify the robust count, include the robust-count statement. The following example configures the robust count to 3.

    The robust count automatically changes certain IGMP message intervals for IGMPv2 and IGMPv3.

    • On a shared network running IGMPv2, when the query router receives an IGMP leave message, it must send an IGMP group query message for a specified number of times. The number of IGMP group query messages sent is determined by the robust count. The interval between query messages is determined by the last member query interval. Also, the IGMPv2 query response interval is multiplied by the robust count to determine the maximum amount of time between the sending of a host query message and receipt of a response from a host.

      For more information about the IGMPv2 robust count, see RFC 2236, Internet Group Management Protocol, Version 2.

    • In IGMPv3 a change of interface state causes the system to immediately transmit a state-change report from that interface. If the state-change report is missed by one or more multicast routers, it is retransmitted. The number of times it is retransmitted is the robust count minus one. In IGMPv3 the robust count is also a factor in determining the group membership interval, the older version querier interval, and the other querier present interval.

      For more information about the IGMPv3 robust count, see RFC 3376, Internet Group Management Protocol, Version 3.

  • You can apply a source-specific multicast (SSM) map to an AMT interface. SSM mapping translates IGMPv1 or IGMPv2 membership reports to an IGMPv3 report, which allows hosts running IGMPv1 or IGMPv2 to participate in SSM until the hosts transition to IGMPv3.

    SSM mapping applies to all group addresses that match the policy, not just those that conform to SSM addressing conventions (232/8 for IPv4).

    In this example, you create a policy to match the group address for translation to IGMPv3. Then you define the SSM map that associates the policy with the source address where these group addresses are found. Finally, you apply the SSM map to all AMT interfaces.

Example: Configuring the AMT Protocol

This example shows how to configure the Automatic Multicast Tunneling (AMT) Protocol to facilitate dynamic multicast connectivity between multicast-enabled networks across islands of unicast-only networks.


Before you begin:


In this example, Host 0 and Host 2 are multicast receivers in a unicast cloud. Their default gateway devices are AMT gateways. R0 and R4 are configured with unicast protocols only. R1, R2, R3, and R5 are configured with PIM multicast. Host 1 is a source in a multicast cloud. R0 and R5 are configured to perform AMT relay. Host 3 and Host 4 are multicast receivers (or sources that are directly connected to receivers). This example shows R1 configured with an AMT relay local address and an anycast prefix as its own loopback address. The example also shows R0 configured with tunnel services enabled.


Figure 2 shows the topology used in this example.

Figure 2: AMT Gateway TopologyAMT Gateway Topology


CLI Quick Configuration

To quickly configure this example, copy the following commands, paste them into a text file, remove any line breaks, change any details necessary to match your network configuration, copy and paste the commands into the CLI at the [edit] hierarchy level, and then enter commit from configuration mode.


Step-by-Step Procedure

The following example requires you to navigate various levels in the configuration hierarchy. For information about navigating the CLI, see Using the CLI Editor in Configuration Mode in the Junos OS CLI User Guide.

To configure the AMT protocol on R1:

  1. Configure AMT tracing operations.

  2. Configure the AMT relay settings.

  3. Configure PIM on R1’s interfaces.

  4. Enable tunnel functionality.

  5. If you are done configuring the device, commit the configuration.


From configuration mode, confirm your configuration by entering the show chassis and show protocols commands. If the output does not display the intended configuration, repeat the instructions in this example to correct the configuration.


To verify the configuration, run the following commands:

  • show amt statistics

  • show amt summary

  • show amt tunnel