Multicast_Pim_sparse_mode
Orhan Ergun 5 Comments

Pim sparse mode – Multicast is used to send the data to the multiple receivers at the same time. Multicast reduces the load on the servers (Senders/Source in multicast term), provides efficient capacity usage on the network links.

multicast flow

Figure – 1 Unicast vs Multicast Flows

Multicast runs on top of UDP.

Multicast uses Class D address range 224.0.0.0/4. RFC 5771 is a Guideline for the IPv4 Multicast Addresses.

Multicast can be used over IP and MPLS.

Multicast routing protocol with PIM can be implemented in two ways; PIM Dense Mode and PIM Sparse Mode.

PIM is protocol independent multicast. It is protocol independent since pim doesn’t require specific protocol. Multicast PIM can use static routing or any dynamic routing protocols for its operation.

Multicast OSPF and DVMRP was the successor of PIM. They were never implemented by the major vendors though. Today most if not all the networks which uses Multicast implement PIM on their network.

Thus understanding different flavor of Multicast PIM protocol is very important. In this article I will explain PIM Sparse mode operation , design comparison between different multicast pim sparse mode methods.

PIM Sparse mode can be used in three ways.

PIM ASM (Any Source Multicast) , PIM SSM (Source Specific Multicast) and PIM Bidir (Bidirectional PIM)

I will first explain all three PIM Sparse mode deployments from the design point of view and then present the design  comparison chart. This article will help you to choose pim sparse mode design option in real life as well as any network design exam including Cisco CCDE.

PIM ASM (PIM Any Source Multicast)

Multicast receiver can receive the traffic from any source/sender in the network. That’s why PIM join message by the router is sent to the all senders. Receivers don’t need to know the sender IP address.

Sender and receiver IP address is always unicast address. Multicast address is only used for the Group information.

In PIM Any Source Multicast routers keep (*,G) multicast entry in the multicast routing table. PIM ASM uses both Shared and Source Specific Tree.

In PIM ASM deployment, there is always Rendezvous point. Rendezvous Point is used for Source/Sender Discovery.

If source would be known as in the case of PIM SSM, then Rendezvous Point wouldn’t be necessary.

When receiver gets the first multicast traffic from the PIM Rendezvous Point, from receiver to the sender, source specific tree is created and multicast traffic is sent directly between sender and receiver without passing Rendezvous Point. But until first multicast packet arrive, multicast traffic flows from sender to receiver through Rendezvous Point.

PIM ASM Scalability : Since until SPT (Shortest Path Tree/Source Specific Tree) switchover from Shared tree, routers keep (*,G) multicast entries which is good for the router memory consumption.

After SPT switchover, since receiver learns the sender, routers start to keep (S,G) entries which consume much more memory compare to (*,G) entries. That’s why PIM ASM compare to PIM SSM is more scalable but less scalable than PIM Bidir.

Application type : PIM ASM is suitable for one sender to many receiver but Rendezvous Point placement can easily create suboptimal traffic flow. Also Rendezvous Point availability should be kept in mind. In PIM ASM RP redundancy is achieved with PIM Anycast RP concept.

Multicast Traffic Flow : Until SPT switchover traffic has to pass through RP, after switchover since shortest path is used between sender and receiver, optimal traffic flow can be achieved. But compare to PIM SSM which uses always shortest path sender and receiver, traffic flow in PIM ASM is less optimal.

Duplicate Multicast Traffic : In PIM ASM, when traffic is switched over to the shortest path tree, the traffic from the rendezvous point is stopped. But some amount of time same multicast traffic is received both via shared tree and shortest path tree. This is not the case with PIM SSM or PIM Bidir since there is no switchover between the multicast trees.

Fast Reroute Support : Multicast PIM routing convergence is depend on the IGP convergence. Whatever routing protocol is used in the network, Multicast is converged after IGP converged in the case of failure. IP FRR mechanisms such as LFA, U-Turn Alternate, MTR-FRR can be implemented for proactive fast reroute. Fast reroute mechanisms can provide 50ms convergence. Also Multicast only FRR can be used to protect only Multicast traffic. Multicast only FRR is not used to protect unicast traffic.

Stuff Experience : Both PIM ASM and PIM SSM is known by many network engineers. The concepts which are used in PIM ASM and PIM SSM, such as Rendezvous Point, Anycast RP, MSDP, IGMPv2, IGMPv3, Loop avoidance with RPF is also known. This is not the case with PIM Bidir. For example Phantom RP for the Rendezvous Point redundancy or specific loop avoidance mechanism of PIM Bidir is not known by many network engineers.

Loop Avoidance : RPF (Reverse Path Forwarding ) is the loop avoidance mechanism which is used by PIM ASM and PIM SSM design. Unicast routing table is used by the downstream nodes to find the senders.

Security : All sources can send all the multicast groups in PIM ASM. Receivers only provides group address information for the intended multicast traffic but any multicast source can send the traffic (Malicious sender can interfere for example) to the groups.

Listen my podcast with Beau Williamson how John Chamber’s presentation was interrupted.

Complexity : PIM ASM requires Rendezvous Point (RP), RP redundancy for high availability, RP placement/engineering for optimal traffic flow and performance impact and understanding shared tree to shortest path tree switchover and so on. Due to all these this is most complex pim sparse mode design option.

Resource Requirement : In the beginning, until SPT switchover it uses shared tree which is good for resource usage since only (*,G) entries are used so not all the source/group pairs are kept in the multicast routing table. But after the SPT switchover since only (S,G) multicast entries are kept, resource requirement increases.

Troubleshooting : Compare to PIM SSM, it is hard since RP , Anycast RP, MSDP is involved. In PIM Source Specific Multicast there is no RP, Anycast RP, MSDP concepts.

PIM SSM (PIM Source Specific Multicast)

Multicast receiver should known the sender/sources. (S,G) multicast entries are kept in the router’s multicast routing table. Always shortest path tree which is also known as source specific tree is used by PIM Source Specific Multicast.

232.0.0.0/8 Multicast address range is reserved and recommended to be used by IANA but doesn’t have to be used.

PIM SSM requires IGMP v3 for (S,G) multicast entries. Source should be known by the receivers.

Receivers can use IGMPv2 but in that case, routers should do the IGMPv2 to IGMPv3 mapping either statically or dynamically through DNS. This feature is known as SSM Mapping.

PIM SSM Scalability : Always (S,G) multicast entries are kept by the routers which consumes more memory and CPU that’s is why least scalable compare to PIM ASM and PIM Bidir.

Application type : PIM SSM is the most suitable pim sparse mode option, for one sender to many receiver application traffic requirement. There is no Rendezvous Point and any RP related concept in PIM SSM.

Multicast Traffic Flow : Always uses Shortest Path Tree (Source Specific Tree). That’s why most optimal traffic flow can be achieved with PIM SSM design.

Duplicate Multicast Traffic : There is no duplicate multicast traffic as in the case of PIM ASM.

Fast Reroute Support : Multicast PIM routing convergence is depend on the IGP convergence. Whatever routing protocol is used in the network, Multicast is converged after IGP converged in the case of failure. IP FRR mechanisms such as LFA, U-Turn Alternate, MTR-FRR can be implemented for proactive fast reroute. Fast reroute mechanisms can provide 50ms convergence. Also Multicast only FRR can be used to protect only Multicast traffic. Multicast only FRR is not used to protect unicast traffic.

Stuff Experience : Both PIM ASM and PIM SSM is known by many network engineers. The concepts which are used in PIM ASM and PIM SSM, such as Rendezvous Point, Anycast RP, MSDP, IGMPv2, IGMPv3, Loop avoidance with RPF is also known. This is not the case with PIM Bidir. For example Phantom RP for the Rendezvous Point redundancy or specific loop avoidance mechanism of PIM Bidir is not known by many network engineers.

Loop Avoidance : RPF (Reverse Path Forwarding ) is the loop avoidance mechanism which is used by PIM ASM and PIM SSM design. Unicast routing table is used by the downstream nodes to find the senders.

Security : Receiver intentionally provides Source and Group information to the network. Only if the requested sources send the multicast traffic to the requested group, then receiver accepts the traffic. That’s why PIM Source Specific Multicast is the most secure PIM sparse mode multicast deployment option.

Complexity : PIM Source Multicast is simplest PIM sparse mode deployment since there is no PIM Rendezvous Point and any related concept. Sources are known , no need source discovery.

Resource Requirement : Since all the routers have to keep all source/group pairs in the network for multicast traffic, PIM Source Specific Multicast is the worst PIM sparse mode design option compare to PIM ASM and PIM Bidir.

Troubleshooting : PIM SSM is the easisest multicast pim sparse mode option from the troubleshooting point of view since there is no extra protocol for source discovery such as Rendezvous Point , MSDP and Anycast RP and so on.

PIM Bidir (Bidirectional PIM)

If the sender will be also receiver then PIM bidir is the most suitable multicast pim sparse mode design option.

PIM Bidir is used for any to any multicast communication. Receiver and the sender/source send and receive multicast traffic at the same time.

There is no shortest path tree in PIM bidir. That’s why always Rendezvous Point is used. Rendezvous Point has to be in the datapath which mean all the multicast traffic always has to go through Rendezvous Point.

RPF check is not used in PIM bidir. Instead loop is avoided by the Designated Forwarder Election mechanism. As per DF election rule, only one router can send the traffic from the multi-access segment.

Although there is Rendezvous Point is used, Anycast RP is not used for the Multicast RP redundancy. Instead Phantom RP concept is used. Anycast RP provides active active load balancing and redundancy for the multicast traffic. Phantom RP is used as active/backup. One RP is used as active and second RP is used as backup and can pass the multicast traffic if the primary RP fails. That’s why PIM Phantom RP doesn’t provide load balancing.

 

You can download the below comparison chart for the PIM Sparse mode design comparison.

 

multicast pim sparse mode

PIM Sparse Mode Design Options

Note : All these information and the comparison chart is taken from the Orhan Ergun’s CCDE Course.

 

 
0.00 avg. rating (0% score) - 0 votes
  • dinka

    it is extremely hard to understand your way of explanation, not clear at all, you basically do not go into any detail, are you sure you understand these concepts or are you just copy pastu=ing from other materials?

  • Evgeny

    Are there mistake in table ? MoFRR with PIM-BiDir…
    MoFRR is a (S,G) feature.
    PIM-BiDir can use diferent RP’s/different paths for different groups though.

    • admin

      Haven’t tried it yet and had a doubt on bidir because of Phantom RP and as you also said Shared-Tree but couldn’t see it in the RFC as well.

  • Pingback: PIM Sparse Mode Vs PIM SSM | Cisco Network Design and Architecture | CCDE Bootcamp | orhanergun.net()