cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
584
Views
5
Helpful
4
Replies

Multiple Randevous points for the same source

Hi,

 

We have a network consist of Nexus core switch 7006 and 4 distribution switches 4500 in different places. The connection between core and distribution is layer 3 EIGRP.

 

We have Cisco IPTV on this network. The source is connected to the Core switch.

And streams should be distributed to the 4 distribution switches.

 

Nexus support only pim sparse mode.

When we configure the RP address to be the core switch, we find delay when when moving from one channel to the next that the stream is asked to be sent through EIGRP so it is expected.

 

I configured one of the Distributions to be the RP and this solved the delay problem for this distribution. But it'll still be suffered by the other distributions.

I see we can configure multiple RP on the network but I discovered that it could be one different randevous point per source. Or the multiple Reandevouz point per the same source will be for high availability.

 

My issue now is that I need to configure multiple Randevouz points for the same source ?

How to acheive this ?

Or is there any way to have PIM dense mode to be supported on my nexus switch to not use Randevouz point ?

 

I'll try to have this source on multiple different Access lists and then will configure a randivous point for each distribution with one of the access lists. Wish it could solve the issue

 

Please advise if you see any hepful action 

 

Thanks

Haitham

4 Replies 4

Francesco Molino
VIP Alumni
VIP Alumni
Hi

Maybe I don't get your concern. Can you give a quick sketch of your network?
When I read your post, I'm thinking about doing a Multicast Anycast RP design. But I want to be sure about your design (including source and receivers) before giving you some configuration hints on that particular design.

Thanks
Francesco
PS: Please don't forget to rate and select as validated answer if this answered your question

We have the source in the DC building with the core and distribution switches

Another 2 building have dist switches

All dists talk EIGRP with the core

RP configured with on all network with DC dist IP address

So all network go to DC dist to get the TV streams

There is no delay When you move between channels on DC dist

But on other 2 dists, it take some time (delay) to get a new channel ( new multicast address from the same source, which mean also joining new group) this delay is because DC dist is the RP and to get join new group on remote dist, it get time to get it via EIGRP

I'm asking how we can provide all the channels ( groups ) to all dists all the time. Not when requested only. You can say to have all dists to get all the group from the source all the time as each of them acting as RP


You want to create a RP on each distribution switches to join the same multicast group?
That's why i was talking about multicast anycast design. You can create the same loopback with same ip across the network that will act as RP. the advantage of that design is that you can have a redundant design. Redistributing it over your igp routing the traffic will reach the nearest RP interface.

You can also tweak eigrp timers to get the best re-convergence time.


Thanks
Francesco
PS: Please don't forget to rate and select as validated answer if this answered your question

You'll find on Cisco some documentation for this kind of design. If you need help let me know.

Thanks
Francesco
PS: Please don't forget to rate and select as validated answer if this answered your question