In the previous lesson, you saw how IGMP gives hosts a way to tell their router which multicast groups they want.
One key question remains: how does that router actually get the stream from the source?What IGMP Already Solved
Thanks to IGMP, your router maintains a list of active group memberships per interface.
It knows exactly which hosts want to receive which groups.
Figure 1 – IGMP: R1 knows its receivers for group 239.1.1.1
In this example, PC1 and PC4 sent IGMP Membership Reports for group 239.1.1.1.
R1 knows exactly which interfaces have interested receivers.
But knowing you have receivers is only half the problem.The Missing Piece
Your last-hop router still has no way to pull the multicast stream from the source through the network.
IGMP does not extend beyond that single hop.
Between your last-hop router and the source, there may be several routers.
None of them know that receivers exist, and none of them are forwarding the stream.
Figure 2 – R1 does not know how to reach the source
This is the gap that PIM (Protocol Independent Multicast) fills.
PIM is the multicast routing protocol that operates between routers to build distribution trees from source to receivers.A distribution tree is the path that multicast traffic follows through the network. PIM does not maintain its own routing table.
Instead, it relies on your existing unicast routing protocol (OSPF, EIGRP, or BGP) to determine the best path toward the source.
This is why it is called "Protocol Independent": PIM works with any unicast routing protocol.
If you switch from OSPF to EIGRP tomorrow, PIM keeps working without any change.Answer the question below
What protocol builds the multicast distribution tree between routers?
Now that you understand the gap PIM fills, let's focus on how it works.
From this point on, we will simplify the topology to a single receiver (PC1) so you can focus on the PIM mechanism without distraction.
The principle is the same for all receivers.PIM Neighbor Discovery
Before any multicast tree can be built, PIM routers must discover each other.
Every PIM-enabled interface sends PIM Hello messages to the multicast address 224.0.0.13 (all PIM routers).
Hellos are sent every 30 seconds by default.
Figure 3 – PIM routers discover each other using PIM Hellos
Through these Hellos, each router learns its PIM neighbors and maintains a neighbor table.
R4 is the first-hop router (closest to the source) and R1 is the last-hop router (closest to the receivers).
This neighbor relationship is the foundation for all PIM signaling.The Sparse Mode Approach
PIM has several modes of operation. The most widely deployed is PIM Sparse Mode (PIM-SM), defined in RFC 7761.
In Sparse Mode, no router receives multicast traffic unless it explicitly requests it.
This is an opt-in model. By default, multicast traffic flows nowhere.This raises a new question. When a receiver joins a group, its last-hop router needs to request the stream.
But where does it send that request? The last-hop router does not know who the source is or where it is located.Answer the question below
In PIM Sparse Mode, what must a router do before receiving multicast traffic?
The Rendezvous Point
PIM Sparse Mode solves this with the Rendezvous Point (RP).
The RP is a router known to all PIM routers in the network.
It serves as a meeting point: receivers join toward the RP, and sources register with the RP.
Figure 4 – PIM Sparse Mode topology with the RP
In this topology, R2 is the RP. Every router in the domain knows that R2 is the RP for multicast groups.
The RP does not need to be the most powerful router. It just needs to be reachable by all PIM routers.Answer the question below
Which router serves as the meeting point for sources and receivers in PIM Sparse Mode?
Your routers now know each other and they know where the RP is. The next step is building the multicast distribution tree so traffic can flow from the source to the receivers. This happens in two stages: the receiver side joins, and the source side registers.
The PIM Join
When your last-hop router (R1) learns through IGMP that PC1 wants to receive group 239.1.1.1, it needs to pull the stream from the network.
R1 does not know where the source is. But it knows where the RP is.
So R1 sends a PIM Join message toward R2 (the RP), using its unicast routing table to determine the correct next hop.
Figure 5 – PIM Join builds the Shared Tree toward the RP
This Join travels hop by hop. Each router along the path adds an entry for the group in its multicast routing table.
The notation for this entry is (*, G), where * means "any source" and G is the group address 239.1.1.1.The result is the Shared Tree, also called the RPT (RP Tree). It is rooted at the RP and shared by all receivers for this group.
At this point, the tree is ready to carry traffic, but the RP has nothing to forward yet. The source has not registered.Answer the question below
What notation identifies a Shared Tree entry in the multicast routing table?
The PIM Register
Meanwhile, the source starts sending multicast traffic to the group address. It has no knowledge of PIM or the RP.
It simply sends packets.
The first-hop router (R4) receives these packets and needs to get them to the RP.
But R4 cannot simply forward them as multicast, because no multicast tree exists yet in the source-to-RP direction.Instead, R4 wraps each multicast packet inside a PIM Register message and sends it as regular unicast traffic directly to the RP's IP address. The Register is a unicast tunnel between R4 and the RP. It uses the normal unicast routing table, so no special multicast path is needed.

Figure 6 – R4 encapsulates multicast traffic in a PIM Register to the RP
The RP receives the Register, decapsulates the original multicast packets, and forwards them down the Shared Tree to all receivers.
The multicast stream now flows end to end, from the source through R4, up to R2 (the RP), and down to R1 and PC1.This is the Shared Tree in action. Notice that all traffic passes through the RP, even though R4 and R1 are connected through R3 as well. The Shared Tree always routes via the RP, regardless of whether a shorter path exists. The RP acts as the central point of control for the group, making it easy to manage multicast distribution in large networks.
Answer the question below
What type of message does the first-hop router send to deliver multicast traffic to the RP?
Answer the question below
Where is the Shared Tree rooted?
Answer the question below
If the R1-R2 link fails, how many routers hold a (*, G) entry for 239.1.1.1?
At this point, your network has a working multicast path: the source sends traffic, R4 registers it with the RP, and the RP pushes it down the Shared Tree to all receivers.
But what prevents a multicast packet from being forwarded in circles?
That is the role of the last mechanism you need to understand.The RPF Check
In your topology, R4 is connected to both R2 and R3. The multicast stream could reach R1 through two different paths.
What stops your router from accepting both copies? Without a loop prevention mechanism, duplicate packets could circle indefinitely.The solution is Reverse Path Forwarding (RPF). When your router receives a multicast packet, it checks the source IP address and looks up its unicast routing table to find which interface leads toward that source.
This interface is called the RPF interface.
Figure 7 – RPF check passes: the packet arrives via R2
If the packet arrived on the RPF interface, your router accepts and forwards it.
If it arrived on any other interface, your router drops it.
In this example, R1's unicast routing table shows that the source 10.1.1.10 is reachable via R2 on interface G0/2.
The multicast packet arrives on G0/2, which matches the RPF interface. The check passes.Answer the question below
What does your router check to determine the RPF interface?
RPF Check Failure
Now consider what happens when the same packet arrives on a different interface.

Figure 8 – RPF check fails: the packet arrives via R3
The same multicast packet arrives on R1, but this time via R3 on interface G0/3.
R1's unicast routing table still says the source is reachable via R2 on G0/2.
The incoming interface does not match the RPF interface.
Your router silently drops the packet.
If the packet did not arrive from the direction of the source, it is either a duplicate or a loop.In the next lesson, you will explore two PIM variants that change the Sparse Mode model: SSM, which eliminates the RP to avoid this single point of convergence, and Bidirectional PIM, which makes the Shared Tree work in both directions.
Answer the question below
What happens to a multicast packet that fails the RPF check?