In the previous lesson, you saw how PIM Sparse Mode builds a Shared Tree through the RP so that multicast traffic can flow from source to receivers.
This model works well, but it has two limitations that matter in specific scenarios.The RP as a Single Point of Convergence
In PIM Sparse Mode, every source must register with the RP.
The first-hop router encapsulates each multicast packet in a PIM Register message and sends it as unicast to the RP.
The RP decapsulates the traffic and pushes it down the Shared Tree.
Figure 1 – PIM-SM: all traffic passes through the RP
All traffic converges at the RP, regardless of whether a more direct path exists between source and receivers.
This design is simple and scalable for most use cases. But two scenarios expose its weaknesses.Two Problems, Two Solutions
The first problem: your receiver already knows the source.
In applications like IPTV or live streaming, the receiver subscribes to a specific source.
Why force the traffic through the RP when your router could build a direct path to the source?The second problem: you have many sources sending to the same group. In a video conference, every participant sends a stream. With PIM-SM, each source requires its own Register message to the RP.

Figure 2 – PIM-SM: each source creates a separate Register tunnel to the RP
10 participants means 10 Register tunnels.
The RP must decapsulate every single one.
PIM defines two variants that address these problems: SSM eliminates the RP entirely, and Bidirectional PIM removes the Register process while keeping the RP.
Answer the question below
In PIM Sparse Mode, what must the first-hop router create for each source?
SSM (RFC 4607) is designed for scenarios where the receiver knows exactly which source it wants.
Instead of relying on the RP to match sources and receivers, SSM lets your router build a tree directly from the source to the receiver.IGMPv3: The Receiver Names the Source
In the previous lessons, you saw IGMPv2 reports where the host tells its router "I want group 239.1.1.1."
The router knows the group, but not the source.
Figure 3 – IGMPv3: PC1 tells R1 both the group and the source
With IGMPv3, the host sends a report that includes both the group address and the source address.
Your last-hop router (R1) now knows exactly who the receiver wants to hear from.Because R1 knows the source, it does not need to ask the RP.
It can build a path directly toward the source using its unicast routing table.
Answer the question below
In SSM, what does the host include in its IGMPv3 report that IGMPv2 does not?
The (S, G) Join
R1 sends a PIM Join (S, G) toward the source.
The S is the source address (10.1.1.10), and G is the group address (239.1.1.1).
This is different from PIM-SM, where the Join was (*, G) and went toward the RP.
In SSM, the Join goes toward the source, not toward the RP.
Figure 4 – PIM Join (S, G): R1 joins toward the source, not the RP
The Join follows the unicast routing table hop by hop.
R1 sends it to R2 (best next hop toward 10.1.1.10), and R2 relays it to R4 (directly connected to the source).
R2 has no RP role in SSM. It simply forwards the Join as a transit router.Each router along the path installs an (S, G) entry in its multicast routing table.
The result is a Shortest Path Tree (SPT) built directly from the source to the receiver.Answer the question below
In SSM, does the PIM Join go toward the RP or toward the source?
Traffic on the Shortest Path Tree
Once the (S, G) tree is built, traffic flows directly from the source through R4, R2, and R1 to PC1.
No Register message is needed. No RP decapsulation. No Shared Tree.
Figure 5 – SSM: traffic flows on the Shortest Path Tree (S, G)
In this topology, the SSM path happens to go through R2, the same router that served as the RP in PIM-SM.
But R2 is just a transit router here.The path is determined by the unicast routing table, not by the RP.
In larger networks, the SPT can take a completely different and shorter path because it is not forced through the RP.SSM Address Range
SSM uses the reserved multicast range 232.0.0.0/8.
Any group address in this range is automatically treated as SSM by your routers.
To summarize, SSM removes three things from PIM-SM: no RP, no Register, and no (*, G) Shared Tree.
It requires IGMPv3 and uses (S, G) entries only.Answer the question below
Which multicast address range is reserved for SSM?
Answer the question below
Which group works as SSM with no RP: 239.232.1.1, 233.0.0.1 or 232.255.0.1?
SSM works when your receiver knows the source.
But what about scenarios with many sources sending to the same group?
In a multicast video conference, every participant sends a stream.
With PIM-SM, the first-hop router must send a separate Register for each source.How do you scale multicast when the number of sources keeps growing?
Bidirectional PIM (RFC 5015) answers this by removing the Register process entirely.A New Topology
To understand Bidirectional PIM, you need a topology where two routers share the same source segment.
This is important because Bidir introduces a new election mechanism that only makes sense when multiple routers are connected to the same network.In this topology, the Source is connected to SW2. Two routers, R4 and R3, are both on the SW2 segment.
R4 is connected to R2 (the RP), and R3 is also connected to R2.
Figure 6 – Bidir topology: R4 and R3 share the same source segment
Both R4 and R3 can hear multicast traffic from the Source on the SW2 segment.
Both are connected to the RP.
This creates a problem: if both routers forward the same packet toward the RP, the RP receives duplicates.Answer the question below
What happens if both R4 and R3 forward the same multicast packet toward the RP?
The Designated Forwarder
Bidirectional PIM solves this with the Designated Forwarder (DF).
On each network segment, the PIM routers elect a single DF.
The DF is the router with the best metric to the RP.
Figure 7 – DF election
Only the DF is allowed to forward multicast traffic up toward the RP on that segment.
R4 has a metric of 5 to the RP, while R3 has a metric of 10.
R4 wins the election and becomes the DF.
Figure 8 – R4 has the best metric to the RP
R3 is not allowed to forward multicast up toward the RP on this segment.
Only R4 (the DF) can forward traffic toward the RP.Traffic Flows Up
When the Source starts sending multicast traffic, both R4 and R3 receive it on the SW2 segment.
But only R4 (the DF) forwards it.
R4 sends the multicast traffic natively toward the RP. There is no Register message and no unicast encapsulation.
The traffic travels as regular multicast, directly up the Shared Tree.
Figure 9 – The designated forwarder sends to the RP
This is the key difference from PIM-SM.
In Sparse Mode, the first-hop router must encapsulate traffic in a Register and send it as unicast.
In Bidir, the traffic goes up the tree as native multicast. No encapsulation, no decapsulation at the RP.The Bidirectional Shared Tree
Once the RP receives the multicast traffic from R4, it pushes it down the Shared Tree to all receivers.
The traffic flows from R2 to R1, then through SW1 to PC1.
Figure 10 – Bidir in action: R4 (DF) forwards, R3 is blocked
The Shared Tree now carries traffic in both directions: up from the source through R4 to the RP, and down from the RP through R1 to PC1.
No PIM Register is needed at any point.
This is why it is called Bidirectional PIM.Answer the question below
In Bidirectional PIM, does the first-hop router need to send a PIM Register?
Answer the question below
Which PIM variant is best suited for many-to-many multicast scenarios like video conferencing?
Answer the question below
If R4's metric to the RP rose to 12, which router becomes the DF?