Multicast Theory Question

From: Eric Phillips (ephillips@squick.cc)
Date: Mon Dec 10 2007 - 14:08:54 ART

Hi All,

I have a more theory-oriented question with regards to sparse-mode PIM. I
understand that when a router receives a multicast packet that it knows of a
RP for it encapsulates that packet inside a PIM Register message and
unicasts that to the RP. The RP then initates a (S,G) SPT back towards the
source, and once the PIM join "chain" reaches the source router the traffic
is sent using the (S,G) SPT. Then the RP sends a PIM Register-Stop message
to the source telling it to stop encapsulating traffic and unicasting it to
the RP.

However, my question is does the source router continue to encapsulate the
traffic after it is also sending the traffic using the (S,G) SPT? If so,
wouldn't that result in duplicated multicast packets because the packets
arrive at the RP via the (S,G) and the PIM Register messages. If the source
router ceases encapsulating traffic in the PIM-Register message once the
(S,G) is completed, why is a PIM-Register-Stop message even needed? I have
read through "Developing IP Multicast Networks," "Troubleshooting IP Routing
Protocols," "CCIE Practical Vol 2," and "Doyle's Routing TCP/IP Vol2," and
none of them answer this question clearly.

Also, none of those sources seem to describe bi-directional PIM or SSM in
any great detail, or describe how to implement and troubleshoot them; but
both are on the blueprint. Does anyone have any good sources for learning
about those things? Or did I miss something in one of those books?

Thanks,

Eric 

-- 
Eric M. Phillips

Senior Network Consultant
LTI Information Technology     http://www.ltiit.com

501 Avis Drive
Ann Arbor, MI  48108
Phone: (734) 929-1400   Fax: (734) 929-1401

This archive was generated by hypermail 2.1.4 : Tue Jan 01 2008 - 12:04:30 ARST