Can PIM-SM operate on a single L3 Switch without a PIM RP

[MarcSims]

Hi,

I know PIM Sparse Mode is always meant to have a RP - but does anyone have experience of a single L3 switch and multicast forwarding where a PIM-RP does not exist?

I have a L3 switch that has multiple VLAN's and I am trying to confirm if a PIM RP is needed for sources and receivers attached to different VLAN's on the same L3 switch. I know PIM will be happier with a PIM RP, and I know about dense mode etc..... but if no PIM RP exists, does the router still create MFIB and MRIB entries and therefore forward traffic between VLAN's, or does the interaction between the MRIB and PIM prevent the state from being created in a PIM-SM environment where no PIM RP exists for the specific multicast group? My gut feeling is that this will work as I expect that the state is created before PIM checks who the RP is - but am struggling to find anything that doesn't simply state "create a PIM RP" or "Use dense-mode". Multicast routing, IGMP and PIM (SM) enabled on the L3 device and it's SVI's. I know this is not best practice but there's always that odd-ball scenario that we have to accommodate - so would appreciate anyone's thoughts or experience on whether this single L3 device will forward the multicast traffic between the multicast enabled VLAN's on the multicast enabled router with no PIM RP for the specific group!

Thanks in advance!

[Harold Ritter]

Hi @MarcSims ,

Yes, configuring sparse mode implies configuring an RP. You can just configure the local L3 switch to be the RP and this will work just fine.

Regards,

[M02@rt37]

Hello @MarcSims 

In PIM Sparse mode the RP is crucial for establishing the shared tree necessary for forwarding multicast traffic, and without an RP, the PIM-SM process is incomplete, preventing the creation of the necessary (*, G) state for multicast forwarding between VLANs.

While the L3 switch will attempt to create state entries based on the sources and receivers, the lack of an RP means that the switch cannot properly build the multicast routing and forwarding tables (MRIB and MFIB), likely resulting in no multicast traffic being forwarded between VLANs. Even though IGMP snooping might allow local forwarding within VLANs, it does not substitute the need for an RP in a PIM-SM environment. Therefore, without an RP, the multicast traffic across VLANs is not likely to be forwarded as expected, and configuring a static RP or switching to PIM Dense Mode, which does not require an RP, would be necessary to ensure proper multicast forwarding in your scenario.

[Giuseppe Larosa]

Hello @MarcSims ,

it is just enough to configure a loopback interface with an IP address and ip pim sparse-mode and then you can add

ip pim rp <loopback-address>

and you have your RP on the box. There are no special requirements like licenses or specific hardware needed.

So you can easily solve this way and it is the recommended way to do it.

Without an RP your multicast can use the ip pim dense mode fallback

ip pim dm-fallback

see

https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst2960xr/software/15-2_5_e/command_reference/b_1525e_consolidated_2960xr_cr/b_1525e_consolidated_2960xr_cr_chapter_011.html#wp2312996432

But the best option is to configure the RP locally as I have explained  above because PIM SM is more efficient.

Hope to help

Giuseppe

[MarcSims]

Thanks to all for your thoughts and replies, your input is appreciated. I understand how simple it is to configure a PIM RP and the role the PIM RP plays in creating state across a L3 environment - I also appreciate the static RP option. I know that Dense mode avoids the need for a RP - the question was will it work - and as M02@rt37 says "likely resulting" - but it is not certain - and this is what I am trying to confirm. If a source starts transmitting to a multicast group without a PIM RP configured, IGMP will do it's thing and I assume the MFIB state will be created (given it is L2 so should have no dependency on PIM) - but obviously just like IGMP, this only relates to within the VLAN. Obviously if a PIM RP were configured, the MRIB state would be created and the register process started - but with no PIM RP, the register process obviously can't start - but if the MRIB state is created before the register process is started, there is a chance that the switch may already have the state created to forward between VLAN's locally - even with the PIM register process failing as a result of no PIM RP for the group. I don't want to get into the specifics of what I am doing and the existing multicast configuration etc - obviously there are several ways to skin the cat (I hope that's not just a UK saying!) - the simplest option for me is simply to allow some groups that are only required on a single L3 switch to operate without a PIM RP configured. Does anyone have any absolute certainty around whether this does or doesn't work based on personal experiences? Thanks all.

[Harold Ritter · Accepted solution]

Hi @MarcSims ,

> the question was will it work

Sparse mode will not work without either a statically or dynamically configured RP.

If the RP is not configured the packets received from the directly connected source will simply be dropped and no (S,G) state will be created.

Regards,

[M02@rt37]

@MarcSims 

The absence of a statically or dynamically configured RP prevents the multicast routing process from functioning as intended.

Without an RP, the switch or router cannot establish the required (*, G) state for multicast groups, and as a result, the (S, G) state, as @Harold Ritter mentioned, which tracks the source-specific forwarding, will not be created either. Consequently, any multicast packets received from the directly connected source will be dropped because the PIM-SM mechanism relies on the RP to initiate the multicast forwarding tree. Thus, multicast traffic will not be forwarded between VLANs, and without an RP, Sparse Mode simply cannot work

[paul driver]

Hello
Just to confirm , having multicast run between routed interfaces using pim sparse-mode without any RP means there can be no initial pim registration of the source and no join registration from the receivers so no MC traffic would begin to flow correctly.

[MarcSims]

M02@rt37 @Harold Ritter - Perfect - thanks. This was just what I was after. I thought there was a chance the L3 forwarding state could be created locally and maintained whilst the PIM process failed to propagate / create state outside of the device. Thanks for confirming this definitely wont be the case - appreciate your support.

[Harold Ritter]

You are very welcome @MarcSims and thanks for the feedback

[MarcSims]

Although..... just had a thought - there are some "default" (background noise) groups in our network that are in use where we do not provide a PIM RP already - so just took a look at one of those - output below:

L3Switch#sh ip mroute 239.255.255.255
IP Multicast Routing Table
Flags: D - Dense, S - Sparse, B - Bidir Group, s - SSM Group, C - Connected,
L - Local, P - Pruned, R - RP-bit set, F - Register flag,
T - SPT-bit set, J - Join SPT, M - MSDP created entry, E - Extranet,
X - Proxy Join Timer Running, A - Candidate for MSDP Advertisement,
U - URD, I - Received Source Specific Host Report,
Z - Multicast Tunnel, z - MDT-data group sender,
Y - Joined MDT-data group, y - Sending to MDT-data group,
G - Received BGP C-Mroute, g - Sent BGP C-Mroute,
N - Received BGP Shared-Tree Prune, n - BGP C-Mroute suppressed,
Q - Received BGP S-A Route, q - Sent BGP S-A Route,
V - RD & Vector, v - Vector, p - PIM Joins on route,
x - VxLAN group, c - PFP-SA cache created entry,
* - determined by Assert
Outgoing interface flags: H - Hardware switched, A - Assert winner, p - PIM Join
Timers: Uptime/Expires
Interface state: Interface, Next-Hop or VCD, State/Mode

(*, 239.255.255.255), 3y42w/00:02:41, RP 0.0.0.0, flags: SJC
Incoming interface: Null, RPF nbr 0.0.0.0
Outgoing interface list:
Vlan318, Forward/Sparse, 33w2d/00:02:11
Vlan412, Forward/Sparse, 32w2d/00:02:38
Vlan411, Forward/Sparse, 18w6d/00:02:41

L3Switch#sh ip pim rp-hash 239.255.255.255
No RP available for this group
L3Switch#sh ip mroute 239.255.255.255 count
Use "show ip mfib count" to get better response time for a large number of mroutes.

IP Multicast Statistics
xx routes using xxx bytes of memory
xx groups, xx average sources per group
Forwarding Counts: Pkt Count/Pkts per second/Avg Pkt Size/Kilobits per second
Other counts: Total/RPF failed/Other drops(OIF-null, rate-limit etc)

Group: 239.255.255.255, Source count: 0, Packets forwarded: 0, Packets received: 4452303
RP-tree: Forwarding: 0/0/0/0, Other: 4452303/7/4452296
L3Switch#

So, although there are no sources at present (or sources not acknowledged with current config), based on this I do think that the *,G is created even when no RP is configured - but I note that mroute count shows that all packets are dropped despite multiple VLAN OIF's showing in the *,G state - so clearly you are right, it does not allow for packets to be forwarded between VLAN's.

Thanks again - appreciate your help with this odd-ball scenario question!

[Harold Ritter]

Hi @MarcSims ,

The *,G is create by the receivers. The S,G will not be created, as the RP is not configured. As I mentioned in a previous post, the lack of an active RP causes the packets received from the source to simply be dropped and S,G to never be created. The dropped packets will cause the "Other drops" counter to increment. That is what you are seeing in the provided output.

Regards,