Multicast Sparse-Dense Mode or Sparse only

[bdube]

Hi everyone,

I'm building a multicast network and one of the question I have is if I should configure PIM interfaces in sparse-dense mode or simply sparse-mode. FYI, i'm planning to use PIMv2 and BSR to elect RP.

With PIMV2/BSR, why do I need to setup PIM i/f in sparse-dense mode as recommended in the config guide.

Also, I understand that using the mixed mode, the leafs routers (the ones connected to receiver's segment) may switch over Shared Tree to Shortest Path Tree if both paths to the source aren't the same. The document recommend to configure a SPT delay to prevent this. Why, just not using SM only. That way, leaf router will never find to real source. Am I right ?

Do we met some problems, or lose something, with configuring PIM sparse mode only?

Thanks for your answer.

Ben

[Giuseppe Larosa]

Hello Ben,

I agree with you: when using Bootstrap the messages travel hop by hop within PIM packets with destination 224.0.0.13.

So with this scenario you can safely deploy ip pim sparse-mode without using the ip pim sparse-dense-mode.

The only requirement is that you are going to deploy multicast groups that are mapped to an RP.

For example it is possible to use anycast RP with BSR but some years I saw configuration examples only using auto-RP.

I tested and it worked fine.

Hope to help

Giuseppe

[Jon Marshall]

Ben

To answer the second part of your post. Even with PIM Sparse mode receiving routers will still create a SPT to the source rather than via the RP (altho the SPT may go through the RP). The idea behind this is to avoid suboptimal paths from source to receiver.

When to switch across to the SPT is not actually specified so Cisco immediately switch across to an SPT as soon as the first packet is received on the Shared Tree.

So whether you use PIM Sparse or PIM Sparse-dense mode SPT's will still be used.

Jon

[bdube]

So, what are the benefits of using RP if leaf routers will switch to SPT anyway?

Thanks for your helpful answer.

Ben

[Giuseppe Larosa]

Hello Ben,

the RP helps to save resources: the SPT will reuse parts/branches of the RPT the shared tree so the RP work isn't useless.

This means saving memory and CPU on the nodes in comparison to setup the SPT everytime from stratch.

Hope to help

Giuseppe