3Com MSR 50 Series, Configuration Manual

Page: 1291 / 2443

How MD-VPN Works 1291
get flooded in the public network, causing bandwidth waste and extra burden on
the PE devices.
To optimize multicast transmission, the MD solution establishes a dedicated
switch-MDT between the PE devices with private network multicast receivers and
multicast sources for any large-traffic private network multicast stream before it
enters the public network. Then, the multicast stream is switched from the
share-MDT to the switch-MDT, to deliver the multicast data to only those receivers
that need it.
The process of share-MDT to switch-MDT switching is as follows:
1
The source-side PE device periodically checks the forwarding rate of the VPN
multicast traffic. Share-MDT to switch-MDT switching takes place only when the
following conditions are both met:
■
The private network multicast data has passed ACL filtering, and
■
The traffic rate of the private network multicast stream has exceeded the
switching threshold and stayed higher than the threshold for a certain length
of time.
2
As in the previous example, PE 1 chooses an idle switch-group address from the
switch-group-pool and sends a switching notification message to all the other PE
devices down the share-MDT. This message contains the private network multicast
source address, the private network multicast group address and the switch-group
address.
3
Each PE device that receives this message checks whether it interfaces with a
private network that has receivers of that VPN multicast stream. If so, it joins the
switch-MDT rooted at PE 1; otherwise, it caches the message and will join the
switch-MDT when it has attached receivers.
4
After sending the switching notification, PE 1 waits a certain length of time and
then starts using the switch-group address to encapsulate the private network
multicast data, so that the multicast data is forwarded down the switch-MDT.
5
After the multicast traffic is switched from the share-MDT to the switch-MDT, PE 1
continues sending switch notifications periodically, so that subsequent PE devices
with attached receivers can join the switch-MDT. When a downstream PE device
has no longer active receivers attached to it, it leaves the switch-MDT.
n
For a given VPN instance, the share-MDT and the switch-MDT are both forwarding
tunnels in the same MD. A share-MDT is uniquely identified by a share-group
address, while a switch-MDT is uniquely identified by a switch-group address.
Each share-group is uniquely associated with a set of switch-group addresses,
namely a switch-group-pool.
Backward Switching from Switch-MDT to Share-MDT
After the private network multicast traffic is switched to the switch-MDT, the
multicast traffic conditions may change and no longer meet the aforesaid
switching criterion. In this case, PE 1, as in the example above, initiates a backward
MDT switching process. When any of the following conditions is met, the
multicast traffic is switched from the switch-MDT back to the share-MDT: