3Com MSR 50 Series Configuration Manual page 1283

Table 49 Basic concepts in MD-VPN
Concept
Switch-group
Switch-group-pool
Multicast distribution
tree (MDT)
Share-multicast
distribution tree
(Share-MDT)
Switch-multicast
distribution tree
(Switch-MDT)
Introduction to MD-VPN
Main points in the implementation of MD-VPN are as follows:
1 The public network of the service provider supports multicast. The PE devices need
to support the public instance and multiple VPN instances, each instance running
PIM independently. Private network multicast traffic between the PE devices and
the CE devices is transmitted on a per-VPN-instance basis, while the public
network multicast traffic between the PE devices and the P devices is transmitted
through the public instance.
2 Logically, an MD defines the transmission range of the multicast traffic of a specific
VPN over the public network; objectively, an MD identifies all the PE devices that
support that VPN in the public network. Different VPN instances correspond to
different MDs. As shown in
instance plane represents an MD, which serves that particular VPN. All the private
network multicast traffic in that VPN is transmitted within that MD.
3 Inside an MD, all the private traffic is transmitted through the MT. The process of
multicast traffic transmission through an MT is as follows: the local PE device
encapsulates the private network data into a public network packet, which is then
forwarded in the public network, and the remote PE device decapsulates the
packet to turn it back into a private packet.
4 An MTI is the entrance to or exit of an MT, equivalent to an entrance to or exit of
an MD. The local PE device sends out private network data through the MTI, and
the remote PE devices receive the private data through the MTI. As shown in
Figure 370, an MD can be thought of as a private data transmission pool, and an
Description
When the multicast traffic of a VPN reaches or exceeds a threshold,
the ingress PE device assigns it an independent multicast address
called switch-group, and notifies the other PE devices that they
should use that address to forward the multicast traffic for that VPN.
This initiates a switchover to the switch-MDT.
Each share-group is assigned a unique switch-group-pool for
switching to the switch-MDT. At MDT switching, an address
(namely, switch-group address) is chosen from the
switch-group-pool. All the multicast packets for the VPN that enter
the public network at a specific PE device are to be encapsulated
using that address. The switch-group-pool of a VPN must not
overlap that of another VPN, and the switch-group-pool of a VPN
must not contain the share-group of another VPN.
An MDT is a multicast distribution tree between all PE devices in the
same VPN. There are two kinds of MDTs: share-MDT and
switch-MDT. Various packets between different MVRFs are
forwarded through the MT constructed by the share-MDT or
switch-MDT. An MDT is automatically created after configuration
and will always exist in the public network, regardless of the
presence of any actual multicast services in the public network or
the VPN.
A share-MDT is an MDT that uses a share-group as it group address.
In a VPN, the share-MDT is uniquely identified by the share-group.
A switch-MDT is an MDT that uses a switch-group as it group
address. At MDT switching, PE devices with receivers downstream
join a switch-group, forming a switch-MDT, along which the ingress
PE forward multicast traffic of a specific VPN in the public network.
Figure 369, the ellipse area in the center of each VPN
Multicast VPN Overview 1283