Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

Anycast RP w/ AutoRP Scaling

Greetings, we are currently in the process of trialing different multicast solutions on our LAN and are keen to increase resiliance where possible, as such we have implemented Anycast RP w/ AutoRP on a seperate switch block for the purposes on this test.

We have applied the following config to 4 distribution switches as per the description below (AutoRP) configuration not included:

Switch A is connected to Switch B and Switch C using routed interfaces

Switch D is connected to Switch  B and Switch C using routed interfaces

Switch A -------- Switch B

      |                      |

      |                      |  

Switch B -------- Switch D

Switch-A - 1.1.1.1

Switch-B - 1.1.1.2

Switch-C - 1.1.1.3

Switch-D - 1.1.1.4

Switch-A

int loopback0

ip address 1.1.1.1 255.255.255.255

!

int loopback1

ip address 2.2.2.2.2 255.255.255.255

ip pim sparse-dense-mode

!

router ospf 1

network 2.2.2.2 0.0.0.0 area 0

!

ip msdp peer 1.1.1.2 connect-source loopback 0

ip msdp peer 1.1.1.3 connect-source loopback 0

ip msdp peer 1.1.1.4 connect-source loopback 0

!

ip msdp mesh-group anycast-rp 1.1.1.2

ip msdp mesh-group anycast-rp 1.1.1.3

ip msdp mesh-group anycast-rp 1.1.1.4

!

ip msdp originator-id loopback 0

ip pim rp-address 2.2.2.2

end

All is functioning correctly and multicast traffic is being routed as expected however i am concerned about how to scale MSDP.

Question #1 - Lets say we have 20 distribution switches and 2 core switches all of which we require to be potential RP's, if i were to roll this out across the LAN would i have to configure all 22 switches as peers under the IP MSDP PEER statement?

For example on every switch:

ip msdp peer 1.1.1.2 connect-source loopback 0

ip msdp peer 1.1.1.3 connect-source loopback 0

ip msdp peer 1.1.1.4 connect-source loopback 0

ip msdp peer 1.1.1.5 connect-source loopback 0

ip msdp peer 1.1.1.6 connect-source loopback 0

- > -> -> ->

ip msdp peer 1.1.1.22 connect-source loopback 0

etc etc etc etc

or

Question #2 - Can i simply peer the distribution switches with the core switches (also MSDP enabled) hence reducing the number of MSDP peers to 2 on the distribution switches but ~22 on the core switches depending on the overall number of connected distribution switches/routers?

In essence, for MSDP to function properly does every MSDP enabled switch/router need to be configured to form a session with every other switch or can it learn SA information from it's directly connected MSDP neigbours? The documentation would suggest that it's similar to an iBGP configuration and that all MSDP routers should be fully meshed?

Any assistance would be most appreciated.

Regards

  • LAN Switching and Routing
Everyone's tags (3)
1 ACCEPTED SOLUTION

Accepted Solutions
VIP Super Bronze

Anycast RP w/ AutoRP Scaling

In most cases MSDP peering matches with you BGP peering. In regards to the number of peering between the Distros and core switches, I don't think you really need to peer the Distros with cores.  Just use the 2 core switches for your MSDP Multicast domain.  The 2 core switches provide redundancy and load balancing. If not, the configuration becomes long and complicated.

HTH

2 REPLIES
VIP Super Bronze

Anycast RP w/ AutoRP Scaling

In most cases MSDP peering matches with you BGP peering. In regards to the number of peering between the Distros and core switches, I don't think you really need to peer the Distros with cores.  Just use the 2 core switches for your MSDP Multicast domain.  The 2 core switches provide redundancy and load balancing. If not, the configuration becomes long and complicated.

HTH

New Member

Anycast RP w/ AutoRP Scaling

Thank you for your reply, looking further into this you make perfect sense, i keep thinking that i need to bring the RP as close to the receiver as possible.but just having the core load balanced and fault tolerant will do the job

Regards

386
Views
0
Helpful
2
Replies
This widget could not be displayed.