Please Help: Multicast deployment in redundant campus LAN

[smart5]

Hi,

May i know how you guys will setup the multicast for 2 pair of 6500s core switches?

Currently for my case is that each pair is connected with a trunk (layer 2) running HSRP to provide redundancy, then these 2 pairs are inter-connected (layer 3) together.

basically most of my end users are connected to access switches connecting to one pair trying to access the windows media server which is connected to another pair of core switches.

Would appreciate if you can share your configuration how i should setup the multicast config. thanks alot.

-------------

http://www.cisco.com/application/pdf/en/us/guest/tech/tk363/c1501/ccmigration_09186a008015e7cc.pdf

With regards to the design guideline provided, i have read and tested the small campus LAN deployment for my other sites, a question in my mind is:

1) is there a need for connectivity between the two loopback interfaces on both core switches? or are they only local significant? Because right now my Core1 is unable to ping to Core2 loopback0 interface which is normal as there isnt a layer 3 path presence to allow such connectivity.

interface Loopback0

description Interface used for RP/MA

ip address 1.1.1.1 255.255.255.255

ip pim sparse-dense-mode

interface Loopback0

description Interface used for RP/MA

ip address 1.1.1.2 255.255.255.255

ip pim sparse-dense-mode

2) For MSDP to works, there must be connectivity between the 2 loopback interfaces?

[ybajpai]

without asking you the reason behind MSDP etc in your network, the short answer is that MSDP requires a tcp connection between the peers. So you do require connectivity between the loopbacks. optionally, you can specify any other interface as the MSDP originator/connect source id that does have connectivity to get ur MSDP session working between the two routers.

if these are not default peers then you also need bgp multicast afi configured.

[smart5]

Thanks. Thats the answer for question 2, what about question 1?

Appreciate some one to help on this. Thanks

[Edison Ortiz]

On Question 1,

Based on the interface description, if those IPs are meant to be used as RP/MA - you must advertise them via a static or dynamic routing protocol.

[smart5]

Hi Edison,

So am i right to say i must put in a static route in Core1 that destine to the RP/MA interface in Core2 through reachable ip address of Core2 management IP?

Its seems that multicast still works even though these 2 interfaces are unreachable.

Can you also help on the 4 core design which is the earlier part of my question? Thanks

[Edison Ortiz]

Multicast still works because you have interfaces running in sparse-dense mode. In the absence of a RP, the traffic will fall back to dense.

If you issue a 'show ip mroute' and see the flags on those routes, you will noticed you have "D" instead of "S".

As for the design, I'm hesitant to give you any recommendation without a formal analysis and this can take quite a bit of time.

[smart5]

I see, so that is the reason why multicast still works. By the way, when i do a 'sh ip pim nei' on core1 it will be able to show the loopback address on core2 which is unreachable from core1, is this normal?

As for the design, is it possible that we communicate through email? because i may not be keeping track on this post. Appreciate if you can email me (smart5@hotmail.com) and i can share with you the network diagram.

Thanks alot.

[Edison Ortiz]

The loopback address won't be shown on a 'show ip pim nei' command. The addresses shown on this output are the physical interfaces connected between the devices, if PIM is enabled.

[smart5]

Seems i may have gotten the wrong command, will update again once im back to office tomorrow.

By the way, i have heard reviews that multicast will stress on the cpu/mem on the switches, is it true? If it is, how do we control it in order to keep the network environment working normally?

[Edison Ortiz]

The command you may be referring to is, 'show ip pim rp mapping'

Multicast may stress the CPU on the switch if it's process switched. How do you change the multicast from hardware switched (default) to software switched ?

1) entering 'no ip mroute-cache' on interfaces

2) placing ACLs against multicast flows

I'm sure there are others but those 2 are the most commons.

[smart5]

Hi Edison,

I'll disable mroute-cache on the interface then. By the way, how do we determine if it is hardware switched or software switched?

Anyway, for my small sites that i configured sparse-dense-mode, i have included static routes to point to individual loopback interfaces. but it seems it doesnt make any differences other then when i do a 'sh ip mcache' it shows some results from the peer core switch.

Can we communicate through email?

[Edison Ortiz]

No, disabling mroute-cache is not recommended. You want multicast packets to run in hardware, not software.

From that small site, can you ping the loopback ?

What do you mean by, 'it doesn't make any differences' ?

For more information on MCAST on Cat6k, please see:

http://www.cisco.com/univercd/cc/td/doc/product/lan/cat6000/122sx/swcg/mcastv4.htm

[smart5]

Remember initially whereby i said pinging to the loopback interfaces of the core switches is not possible?

I have include static route on both core switches so that both of them can ping each other's loopback interface.

It doesn't have any difference in multicast functionality even when its not pingable or pingable.