MSDP

Hello GS,

I am doing the IE Multicasting Technology Labs v4.1 at the moment. The MSDP
Lab in particular on page 74 and I need some help.

I have the entire lab setup and running MSDP and PIM-SM between the
interfaces on R2 and R4, to really test this out I have joined SW1 and SW2
to the igmp group 224.1.1.1 and icmp'd the group address from R5. I get
replies from both switches and everything works just dandy.T

Then to perform some further testing and to make sure it was definately
working correctly I took of the MSDP configuration, cleared all the
multicast routes and tested the scenario again, believing that I would only
get replies from SW2 which is in my local AS but this was not the case,
because there was a PIM-SM neighbor relationship between R2 and R4, R2 sent
a join message to R4 and look out I'm up and running between two domains
without MSDP.

Ok, so I figured that was silly, I should turn PIM-SM off the interfaces
between R2 and R4 then MSDP back on and then re-run both tests. I did that
and now i'm even more confused, the behaviour I got was when I sent through
the first ICMP to the group it went through and I got a reply from both
neighbors, after that the PIM-SM behaviour where the Receiver will create a
new SPT straight to the source was created instead of using the RP; because
of this behaviour the RP decides that there are no recipients for the stream
coming through him anymore and sends a prune message, to which none ot the
multicast traffic is sent to R5 and therefore never forwarded across the
domain!

So my query here is, what is the point of MSDP for interdomain connectivity
like this presented on the lab for interdomain connectivity if it works with
PIM-SM without it... or is there something I'm missing with my MSDP
configuration to get around PIM-SM RP pruning itself from the tree after the
first ICMP?

I haven't worked my way up to multicast BGP but would this get around the
issue or would I have the same problems??

Thanks in advance for your help GS!

Regards,

Garth

Blogs and organic groups at http://www.ccie.net
Received on Sun May 17 2009 - 12:46:27 ART