Multicast - PIM dense vs sparse mode

When deploying multicast in a network topology, multicast routing must be employed to ensure the correct forwarding of multicast traffic. In particular, the multicast routing protocol that is most often used is Protocol Independent Multicast (PIM). There are two types PIM that can be employed:

In PIM dense mode, the multicast traffic is initially flooded to all parts of the network using a construct called the shortest path tree (SPT). Parts of the network that don't want or hasn't requested multicast will prune back the traffic. In this sense, dense mode is a “push” model. PIM dense mode primarily uses the SPT, with each source having a separate distribution tree.

PIM sparse mode initially sends no traffic until a receiver on a network segment indicates it is interested in receiving the traffic. The traffic is then sent to that network segment by the RP (Rendezvous Point) in a shared tree topology. This is called a “pull” model. PIM sparse mode typically uses the shared tree topology, most often called a Root Point Tree (RPT) initially. This delivers greater simplicity of implementation and lower overhead. When a host wants to receive the multicast traffic, it sends an IGMP join message upstream toward the RP. Once traffic for a group is flowing, if the volume of traffic and the number of receivers warrant it, PIM sparse mode can switch over to the SPT for greater efficiency.

There is a third option called PIM sparse dense mode in which sparse or dense mode is used on a per multicast group basis.