cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
8331
Views
46
Helpful
7
Replies

Can someone explain shortest path switch over in Multicast?

markmarek
Level 1
Level 1

Can someone explain shortest path switch overin Multicast?

1 Accepted Solution

Accepted Solutions

arun kumar
Level 1
Level 1

Hi,

Shortest path switchover is used in PIM Sparse mode.

The concept of Sparse Mode is the multicast traffic should only be delivered when there are active receivers in the network. To accomplish this SM has explicit Join and Prune mechanism rather than Flood and Prune used by Dense mode.

When the last hop router (where receiver is connected) receives an IGMP join message, it sends a PIM Join towards RP. This creates shared tree from the last hop router till RP along the routers.Thus a shared tree has been created from last hop router till RP.

At the same time the above procedure was taking place, the first hop router (where source is connected) starts sending PIM Register messages to RP (when source starts sending multicast traffic). This register messages are unicast so that the intermediate routers will not aware of multicast traffic. When the RP receives Register messages from first hop rotuer, it does below checks after decapsulation:

1. Are there any shared tree for the specific multicast group. If yes then send a SPT join towards first hop router so that the multicast traffic from source will be delivered to RP via native multicast rather than encapsualted packets. Also RP sends Register Stop message to inform first hop router to stop sending encapsulated multicast packets.

2. When there are no active groups in RP (i.e RP has not received any PIM join for the multicast group from any last hop router), RP sends a Register Stop message.

By this the traffic from source flows via SPT till RP. From RP it flows via shared tree till receiver. Everything is fine till now. But the drawback in SM is placement of RP and the load on it. Since RP is located in a central location chances are there that the multicast traffic from source takes a longer path to reach the receiver. This will result in increase in latency along with increase in RP load.

So in those cases, it is desirable for the multicast traffic from source flows to receiver on optimal path. This is done by SPT switchover and this is done when the multicast group traffic crosses the SPT threshold configured. In Cisco the default SPT threshold is Zero kbps. So when the last hop router receives first multicast packet via shared tree, it does SPT switchover. This is done by sending a SPT Join towards the source thus bypassing the RP. Now the traffic from source prefers the most optimal path to reach the receiver.

The same time, last hop router sends a PIM prune towards RP to inform it that it does not want multicast traffic over shared tree. This is done to avoid duplication of the multicast packets over SPT and shared tree.

Few facts:

1. Last hop router needs information about the source to send SPT Join and this can only happen when it receives few initial multicast packets via RP shared tree.

2. If you do not want the last hop router to fall back to SPT switchover, then configure SPT threshold to infinity in last hop router.

Sorry for the long post.

HTH.

thanks

Arun

Pls rate if it is helpful

View solution in original post

7 Replies 7

arun kumar
Level 1
Level 1

Hi,

Shortest path switchover is used in PIM Sparse mode.

The concept of Sparse Mode is the multicast traffic should only be delivered when there are active receivers in the network. To accomplish this SM has explicit Join and Prune mechanism rather than Flood and Prune used by Dense mode.

When the last hop router (where receiver is connected) receives an IGMP join message, it sends a PIM Join towards RP. This creates shared tree from the last hop router till RP along the routers.Thus a shared tree has been created from last hop router till RP.

At the same time the above procedure was taking place, the first hop router (where source is connected) starts sending PIM Register messages to RP (when source starts sending multicast traffic). This register messages are unicast so that the intermediate routers will not aware of multicast traffic. When the RP receives Register messages from first hop rotuer, it does below checks after decapsulation:

1. Are there any shared tree for the specific multicast group. If yes then send a SPT join towards first hop router so that the multicast traffic from source will be delivered to RP via native multicast rather than encapsualted packets. Also RP sends Register Stop message to inform first hop router to stop sending encapsulated multicast packets.

2. When there are no active groups in RP (i.e RP has not received any PIM join for the multicast group from any last hop router), RP sends a Register Stop message.

By this the traffic from source flows via SPT till RP. From RP it flows via shared tree till receiver. Everything is fine till now. But the drawback in SM is placement of RP and the load on it. Since RP is located in a central location chances are there that the multicast traffic from source takes a longer path to reach the receiver. This will result in increase in latency along with increase in RP load.

So in those cases, it is desirable for the multicast traffic from source flows to receiver on optimal path. This is done by SPT switchover and this is done when the multicast group traffic crosses the SPT threshold configured. In Cisco the default SPT threshold is Zero kbps. So when the last hop router receives first multicast packet via shared tree, it does SPT switchover. This is done by sending a SPT Join towards the source thus bypassing the RP. Now the traffic from source prefers the most optimal path to reach the receiver.

The same time, last hop router sends a PIM prune towards RP to inform it that it does not want multicast traffic over shared tree. This is done to avoid duplication of the multicast packets over SPT and shared tree.

Few facts:

1. Last hop router needs information about the source to send SPT Join and this can only happen when it receives few initial multicast packets via RP shared tree.

2. If you do not want the last hop router to fall back to SPT switchover, then configure SPT threshold to infinity in last hop router.

Sorry for the long post.

HTH.

thanks

Arun

Pls rate if it is helpful

Very well explained...

Long post but very good stuff

rated as it deserves

Best Regards

Giuseppe

Thank you!!!!! I am doing my labs and your explanation really help me

Nice topic ! Very helpful

nathanieljhnsn4
Level 1
Level 1

Excellent write-up Arun!

Thanks great post.

 

My question is. If the last hop router dont have a rpf route back towards the source but have a rpf route to the the RP will the router stay on the shared tree and forward multicast packets.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Review Cisco Networking products for a $25 gift card