The working principle of the multicast Dr and the detailed explanation of the troubleshooting

1, the problem description:

We have a CPE MP1803 router as a customer's CE router, the PC after the IGMP report, our router will be in the IGMP table to write the multicast group, but the customer said that the upstream of the PE device does not receive the PIM of our equipment Join message and ultimately cannot draw the multicast traffic down.

After the investigation, found that customers in the same local area network has multiple CPE, and our MP1803 is not Dr.

So that's why the client has failed to catch the PIM join message on the upstream interface of our router after debug, which is mistaken for the problem of our router.

At that time, the client was advised to disconnect the LAN, and then directly with the PC to MP1803 LAN port above, this time the PC multicast normal work.

Problem found. The following is about the work of the LAN Dr Principle of discussion and experiment.

2, the working principle of multicast Dr.

PS: Before looking at the document, you need to recall how IGMP works, and if the multicast client joins the common principles section of the RP.  Detailed details can refer to the documentation: a very strong CCIE note. CISCO CCIE Routing and Switching–kaka ' s note

Author: Fang Zhiyong

Personal Understanding:

Now my understanding of DR is that if a LAN has multiple routers at the same time, then in show ip Pim neighbor, for the LAN will also elect a Dr out.

When the client sends an IGMP message, it has nothing to do with the PC's default gateway, and the IGMP answer is sent to all upstream routers, and the PIM join is not encapsulated until Dr receives a downstream PC's IGMP Message sent to RP to join. For another non-Dr device, when he receives the IGMP address, he will only record the anchor group in his IGMP table, but he will not send the PIM join message to the RP, so if someone says this route show IP PIM Mroute Group x.x.x.x state is not joined, then you need to see whether the router is a DR to the LAN. If it is not Dr then this phenomenon is normal and should be the case.

And if this router sends a PIM join, the show IP Pim mroute on this device should be able to find the (*,G) table entry for the desired multicast group, and then the upstream router, and there should be (*,G). Because we sent it to the top. If we have the above (*,g), then we should check if the router multicast is working properly.

Here again a problem, in the same LAN, only Dr Forwarding anchor traffic, only to send PIM join message, then if the host group more cases, the local area network Dr is also a performance test?

That must be true.

So in the design of the network, if the local area network host group in great demand, then the CPE router must use a higher performance equipment to do, otherwise it is likely because of performance bottlenecks.

My recommendations for the Agreement:

Mp1800-d (config-if-fastethernet0) #ip PIM dr-priority 200? <CR>

Mp1800-d (config-if-fastethernet0) #ip PIM dr-priority 200

Q: You see here are all based on interfaces, but I would like to 224.1.1.1 R1 as a dr,224.1.1.2 R2 as a Dr, to achieve a load balancing function, this can be done?

A: I think really is a demand, if I have 20 of the host group need to run, each is high-definition video streaming, it is not a high demand for CPE to do it? Each HD according to 4M to calculate, are 100M HD .... Originally 4 1800 can be, now must be 3 MP1800 plus a 2824 to do this thing, and can not do the backup. In other words, it takes 4 2824 to fully hot standby, or the main 2824 is paralyzed, then any 1800 will be paralysed because a 1800 can not achieve 100M HD video traffic in line speed forwarding.

3, CCIE Learning notes on the above software module description. (about the key I marked with red)---the following are the principles of Cisco, command output may be slightly different, the same principle, you can refer to.

The above is the message format of the PIM join message.