IGMP Version 3


(Rene Molenaar) #1

This topic is to discuss the following lesson:


(hyun p) #2
It seems like you omitted "no" before the command, ip igmp join-group 239.1.1.1 source 1.1.1.1.

What happens when H1 leaves a group? Let’s find out:
H1(config)#interface Gigabit 0/1  
H1(config-if)#ip igmp join-group 239.1.1.1 source 1.1.1.1

(Rene Molenaar) #3

Hi Hyun,

Thanks for letting me know, just fixed it.

Rene


(Abhishek D) #4

liked the fav video :slight_smile:

a. so this 1.1.1.1 is the IP address of video server ?
b. in today’s scenerio which igmp version is mostly used now ? is it v3 ?
c. is it correct that here HOST will be actully a router behind which we have PCs ?
d. wont it cause more over-head on client side as now they will have to configure specific IP of video server? is it feasible from scale point of view ?

thanks


(Rene Molenaar) #5

Hi Abhishek,

Yes, 1.1.1.1 is the IP address of the multicast source, that could be a video server. IGMP Version 2 is probably still the most common version.

The location of your sources / hosts really depends on your network design. In a small network, it’s possible that you only see switches where the source/hosts are in the same L2 design. In larger networks, your hosts are on the access layer and the source(s) might be somewhere else, behind some router(s).

The client has to keep track of the multicast groups that it wants to receive, each client only joins 1 (or a few) multicast group(s) so there’s not much overhead there.

Rene


(AZM U) #6

Hello Rene,
A few questions and confusions.

  1. When a multicast address is being used as a group address(for example 239.1.1.1), this address is assigned to the multicast server and this address has to have route in the network design because whenever a host will encapsulate an IP packet, it will use its own address as the source and 239.1.1.1 address as the destination address. Is this correct?
  2. Would you also please give me a real life scenarion where multicast is used?
  3. In your IGMP version 3 example, 1.1.1.1 is being used as the source address. What is this address for? As far as my understanding goes, source IP address is the host IP and destination is the Multicast IP address(239.1.1.1). Please explain it little bit.
  4. One difference between ver 2 and ver 1 is that in ver 1 all the hosts send report to the router whereas in ver 2 only host sends report to the router on behalf of all the hosts in the same group. Is it correct?

Thank you so much

Azm


(Lazaros Agapides) #7

Hello Azm

Well, not quite. The multicast address is not assigned to the multicast server, but it is the destination used for multicast traffic FROM the server TO the members of the group. Secondly, under what circumstances would a host communicate with 239.1.1.1? When it makes a request to join the multicast group. This in essence is a communication between the host and the IGMP functionality of the router, not with the multicast server. This communication would have a source IP of the host and a destination IP of the mutlicast IP, but that destination in such a communication is the router.

* Videoconferencing
* IPTV/Digital TV/Cable TV
* Live video transmission over the Internet
* Webinars
* online radio

Here Rene is indicating that a host can also include a source IP address in the membership report messages and this address can be configured. Because the host is a router, it can state which one of its IP addresses is the source for the specific multicast transaction. Rene has added a fictitious IP address just to see it easily in the debugging and wireshark captures. In a production environment, this IP address should be that address that will be added to the multicast group.[quote=“azmuddincisco, post:6, topic:1123”]
4) One difference between ver 2 and ver 1 is that in ver 1 all the hosts send report to the router whereas in ver 2 only host sends report to the router on behalf of all the hosts in the same group. Is it correct?
[/quote]

I’m not sure I understand what you mean. Can you clarify the second part of your question?

I hope this has been helpful!

Laz


(AZM U) #8

Hello Laz,
Thanks a lot for your time explaining it. However, I did not quite get it and you may need to explain it little bit more to me. In order for the explanation, I like to have a scenario.

Let’s say I am sitting in a small office network and my computer(10.10.10.10) is connected to the internet. I am trying to tune 100.100 FM radio sitting in the internet cloud and the IP address of the FM server is 1.1.1.1. Would you please describe the packet flow here while I am trying to connect to the FM radio? Thank you so much.

Azm


(Rene Molenaar) #9

Hi Azm,

Multicast isn’t really used on the Internet. It would make a lot of sense to use multicast for online radio but in reality, it’s unicasted everywhere. There is no “global” multicast network that spans multiple ISPs. One application that was common back in the days on the LAN was Norton Ghost to send a system image to all computers. Without multicast, you had to unicast huge images to all computers which wasn’t very efficient.

When your computer wants to receive a multicast stream, it uses IGMP to “report” which multicast group it wants to receive. IGMP version 2 is very common but it doesn’t allow you to specify a source which means that everyone can send packets to 239.1.1.1. With IGMP version 3, you can specify the source which means you can report that you only want to receive traffic for multicast group 239.1.1.1 from source x.x.x.x.

Once a multicast enabled router receives your IGMP packet, it’s up to PIM. It really depends if you are using PIM dense or PIM sparse mode how the multicast traffic is flooded in the network. PIM dense uses a “flood and prune” model while PIM sparse uses a “join” mechanism to request the multicast traffic from an upstream router. These two lessons explain that process:


(AZM U) #10

Thank you so much Rene.

Azm


(Hussein Samir) #11

Hi Rene,

What is the purpose of sending two membership report from H1, one determines there is a new address and the other determines the mode, I mean why not be in one membership report ???


(Rene Molenaar) #12

Hi Hussein,

To be honest, I have no idea…I guess this is one of those “that’s how they designed it” answers. They probably could have been combined in a single packet but for some reason, they decided to go for two packets.


(Hussein Samir) #13

Thanks Rene I appreciate your efforts and thank you for not falling short


(zhenhua m) #14

Hello Rene,
I am from China and my English is not very well. I just hope you add a video to each lesson. By the way, may i know if you have recorded videos from CCNA to CCIE. Thanks !


(Lazaros Agapides) #15

Hello Zhenhua

It’s great to have you here! Rene has created videos for various lessons. All of the new lessons that he creates have videos, but I’m not sure if he will eventually add videos to the older lessons. Many of the lessons in CCNA, CCNP and CCIE have videos, but not all. I hope you find the videos as well as the forum useful!

I hope this has been helpful!

Laz


(Dmitriy G) #16

Hi Rene,

What will have if H1 joins 239.1.1.1 group and includes source 1.1.1.1 and H2 joins 239.1.1.1 group and excludes source 1.1.1.1 ?

Thanks,
Dmitriy


(Heng S) #17

Hi Rene,
On Host no need to specific igmp version 3 ? as by default it is version 2.


(Lazaros Agapides) #18

Hello Dmitriy

If H1 sends a membership report message stating that it wants to join 239.1.1.1 and it includes the extra piece of information that states that it will only accept multicast streams from 1.1.1.1. Whenever 1.1.1.1 sends traffic to 239.1.1.1, it will go via R1, which will check the source IP address and see that it is 1.1.1.1. It is acceptable to H1, so R1 will stream multicast traffic to H1.

H2 however, has sent a membership report message to R1 joining 239.1.1.1 but excluding 1.1.1.1. This means that when 1.1.1.1 sends multicast traffic, R1 will receive this and will NOT send any of the traffic to H2, because the source is 1.1.1.1 and this has been excluded.

I hope this has been helpful!

Laz


(Lazaros Agapides) #19

Hello Heng

You are correct, that we must change the IGMP version to version 3 because the default version is 2. This is stated in the lesson as well:

Also, you can see in the configuration of the interface, that version 3 has been enabled:
image

I hope this has been helpful!

Laz


(Heng S) #20

HI Laz
Yes I see , on R1 we change igmp version to version 3. But how about H1 ? i don’t see you change it in the lesson.
And another question, can you explain about RP (Rendezvous Point) ?. I am really confuse about it.