Redundant RPs with identical IP addreses?

[fsebera]

Topology Overview

We have two hub sites with about 300 spokes in a true “hub and spoke” MPLS topology.  Half of the spokes use a single hub-site for services while the other half of the spokes use the opposite hub-site for services.  There is a backup link - isolated from MPLS cloud - between the two hub-sites as-well.  Everything works very well at this point. 

The PLAN

We want the hub sites to provide redundant multicast rendezvous points (RP) to the spokes.  We use pim sparse-mode only with auto-rp and mapping agents; totally dynamic. Currently, we will only source from the hub sites (we don’t currently plan to source from any spoke).  We will not simultaneously source from the two hub sites; solo sourcing only.  Nor will we provide Video-on-Demand or VTC.  We plan to source HQ updated information on a continuous loop out to the spokes to allow the end-users to “tune-in” when convenient; sort of like a radio station or a broken record.

The Question

Since we are using dynamic RP dissemination, is there any advantage to making the Rendezvous Points addresses the same or different?

Eg.

   Hub-Site-1

interface Loopback99

  ip address 10.99.99.99/32

  ip pim sparse-mode

  Hub-Site-2

interface Loopback99

  ip address 10.99.99.99/32

  ip pim sparse-mode

Tks!! Frank

[fsebera]

Where ol where are all the multicast folks¿?

Update:

I am currently configuring the RPs with duplicate addresses and have noticed a few issues.

1. The MPLS PE "aggregation" router chooses a single prefix and thus forces ALL traffic destine to the RP address through a single hub regardless of the spoke source (not multicast source).

I have the prefix 10.99.99.99/32 arriving at the PE router that supports 2 spokes (hub and spoke design), one 10.99.99.99/32 came from hub1 while the other 10.99.99.99/32 came from hub2.  BGP filters through the list of route selection parameters and determined the oldest prefix is the one to select as best.  Now both spokes route using this single prefix. This was easily fixed with MPLS route-targets and LP assisgnments!!

2. I now have 2 active RPs (single source) in my single multicast domain which violate basic multicast rules.

I implemented MSDP to sync up the two RPs to avoid RPF failures.

3. Failover is instant as the spokes RP mappings always have the RP address 10.99.99.99 regardless of where it came from. If hub1 fails, hub2 is still providing the same RP address and it's already in the RP mapping table.

ok, hopefully someone ---with more multicast experience than myself ---will jump in and assist me  - P L E A S E!!!!!!!

Tks

Frank

[redemptionlight]

you cannot have same identical ip addresss for all interface in your network acting as RPs. it will only cost replicate addresses. i tried that and its not efficient.......

use different ip addresses. they could be in the same subnets if you like.......

************************************************************************************************************************************

PS

I have a Multicast network topology shown below

http://dl.dropbox.com/u/20145606/multicast_network.png

I have implemented multicast on the network in a hub and spoke topology. i have set up ospf routing protocol and broadcast all network and can successfully ping.

I am currently using VLC player as my media streaming server and client. i have set up rtp streaming from the HUb router using multicast ip 224.2.2.2 and unable to broadcast the multicast traffic across the spokes 1 and 2 PC's

I have never used vlc player  never set up multicast network before and i am struggling with this and need help.

these are my router configs below

http://dl.dropbox.com/u/20145606/ip%20video%20config.txt