In that event, you may want to look into "ip multicast multipath" .
 
Remember balancing is done per flow on routers.
 
Check out "debug ip mpacket" for actual RPF behavior on your setup.
 
 
Scott Morris, CCIE4 (R&S/ISP-Dial/Security/Service Provider) #4713, JNCIE
#153, CISSP, et al.
CCSI/JNCI-M/JNCI-J
VP - Technical Training - IPexpert, Inc.
IPexpert Sr. Technical Instructor
 
A Cisco Learning Partner - We Accept Learning Credits!
 
[EMAIL PROTECTED]
 
Telephone: +1.810.326.1444 
Fax: +1.810.454.0130
http://www.ipexpert.com

  _____  

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Tim Riegert
Sent: Monday, June 04, 2007 10:24 PM
To: [email protected]
Subject: [OSL | CCIE_RS] PIM RPF Check


Sorry if this is a rudimentary question...
 
 
When using PIM-DM, if a router performs an RPF check to prevent looping
multicast packets by looking at the unicast routing table to determine the
interface it would send to reach the source, what happens if the router has
multiple equal cost routes to the source?  Wouldn't the router receive the
packets on all those interfaces when it is flooded, and each packet would
pass the RPF check, and then the router would flood it back out the other
interfaces.  Obviously, this behaviour would stop when the packet reaches
the upstream routers again, but it just seems like an inefficient check. 
 
For instance, dual core design connecting to two building distribution
routers (meshed) - host in building A sources a packet destined for a
multicast group. Building A router floods packets to both cores - each
packet passes RPF check.  Both cores then send the packet to building B
router (which has equal cost paths to the two cores).  Router in building B
then performs RPF check on both packets, and they both pass?  Or does the
router choose one route from the unicast routing table as the "best" for
PIM?  If both packets pass, then Building B router forwards the packet that
it received from core1 to core2, which then fails RPF, and building B router
also forwards packet received from core2 to core1, which then fails RFP on
core1?  I know that PIM-DM is inefficient, but that would seem to be a
really bad design in a large campus. 
 
THanks in advance.

Reply via email to