• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

igmp snooping - sender & receiver on same vlan

C

Cooky

Golden Member
Background:
Before whenever our Desktop group used Altiris to image PC's & laptops, it became broadcast because no PIM RP was available to facilitate the register process (I think), even though the sender & receivers are all in the same vlan.

So we simply implemented PIM routing on the core L3 3560 to enable multicast.

Question1:
Would the image that the source sends have to traverse through the L3 3560 (PIM router), and then back down the tree to reach the receivers?

Or would the multicast stream propagate through the access / distribution switches, w/o having to go through the 3560 first?
All the switches have IGMP snooping enabled.

Most of the multicast documents I've read seem to only cover forwarding scenarios based on L3 PIM routers, and not L2 switches.

While I have a basic understanding on how multicast forwarding & IGMP snooping work, I'm clueless on how the multicast stream looks like on a LAN.

Question2:
Follow-up to Q1, if the sender & receivers happen to be on the same L2 access switch, would the multicast stream reach the receivers directly, or would the stream go up to the PIM router (L3 3560), and then back down to the L2 switch?

Question3:
How would the scenario change if sender & receivers are on different subnet/vlan? (assuming PIM is enabled on all SVI's)
Would an access switch have different snooping tables for each vlan?
Or would multicast stream have to be PIM routed by the L3 switch?
 
Last edited:
IGMP snooping is great. The switches learn who the receivers are and in what groups by monitoring the join/leaves. This is all working at layer2 so it doesn't matter if they are on the same subnet or not, it just determines what DST ports the switch is to flood a MCAST stream. Very little configuration work, just turn it on (on by default actually). If on same broadcast domain, then the stream won't go to the router unless the router joins the group (and would only do so if the router had receivers and a mroute to those receivers.

Once you cross a L3 boundary that's where PIM and multicast routing come into play, this is working at L3 to build the mcast tree for a specific MCAST address.
 
spidey, as always, appreciate the prompt response & the insight.

So, it sounds like after the RP register & SPT switch-over, the access L2 switch would just forward stream to receivers directly until they send IGMP-leave?

I'm a bit confused by your statement:"If on same broadcast domain, then the stream won't go to the router unless the router joins the group (and would only do so if the router had receivers and a mroute to those receivers."

When receivers want to start getting stream, they'd send IGMP-join up to the router, so the router should have mroute & the receivers' SVI under outgoing interface.
Therefore, stream should still go through the router, assuming it's the only router on network (so doesn't matter if it's the RP & if it's in path after SPT-switchover), right?
Or does the router only handle IGMP join/leave at that piont? (still under assumption sender & receiver in same subnet/broadcast domain)

The tech at our remote site told me after I enabled multicast, imaging actually slowed down slightly...not sure why that would be.
Will have to do more testing.

==
Here's the mroute output on our L3 switch:
(*, 239.193.0.11), 1d02h/00:01:59, RP 0.0.0.0, flags: SJC
Incoming interface: Null, RPF nbr 0.0.0.0
Outgoing interface list:
Vlan176, Forward/Sparse, 22:45:05/00:01:59
 
If the sender and receiver are on same vlan, igmp snooping is used to determine what ports to flood the stream based on the igmp joins. A router isn't even involved in this case.

If the router does have a mroute and out going interface is the same as the source then the router still isn't involved, why would it need to be? Follow me?

If, and only if, the router has receivers on a subnet different from the source is the mroute actually populated with the specific source (otherwise you'll see a (*,MCAST-ADDY). In this case then of course the router is involved. Make sure you use private mcast addresses to make sure you've not sending to clients that don't need to receive it. 239.x.x.x/8
 
ok makes sense now.
thanks for the explanation, and yes, we do use the private range only (239.0.0.0).
 
simply put, IGMP snooping makes a L2 switch multicast aware by looking at the IGMP messages. The switch will only forward ethernet frames to the hosts that want to join a certain multicast group. L2 switches that don't have this functionality will treat muticast the same as broadcast traffic and flood all traffic. You can never break something by enabling it (on the contrary I would say)
 
freegeeks said:
simply put, IGMP snooping makes a L2 switch multicast aware by looking at the IGMP messages. The switch will only forward ethernet frames to the hosts that want to join a certain multicast group. L2 switches that don't have this functionality will treat muticast the same as broadcast traffic and flood all traffic. You can never break something by enabling it (on the contrary I would say)
Click to expand...

Unless the switches support CGMP.:sneaky:
 
freegeeks said:
hehe, yeah

well CGMP is cisco proprietary and generally sucks compared to IGMP snooping
Click to expand...

I know, just being a smart ass.

IGMP snooping is indeed a wonderful thing, probably why it's enabled by default. Flooding those image streams at 1 Gbs/s to every single wiring closet is not a good idea.
 
spidey07 said:
I know, just being a smart ass.

IGMP snooping is indeed a wonderful thing, probably why it's enabled by default.
Click to expand...

I know you were joking, I'm not heavy into Cisco anymore (been working on Alcatel-Lucent SR 7750 mpls stuff the last 2 years) but I noticed that on some Catalyst they dropped cgmp (could be wrong though...)
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top