I’ve managed to kind-of-sort-of get multicast working in a lab environment but it seems buggy. Like i’ll have a working configuration and then i’ll reboot a router and all of a sudden, stops working
Or I try and break it on purpose, and yet multicast is still flowing
I’m just confused with it all, so i’m going to throw this out there and hopefully someone can clear it up for me
The scenario is this. We have a fully routed MPLS + OSPF network. We have a customer wanting for us to deploy a multicast setup for his sites because he has IP camera streams
At the moment all sites connect via PPPoE connections, over VPLS tunnels. I don’t want to use those PPPoE tunnels but thought it worth mentioning incase VPLS has any influence here. My plan is to give out another VLAN for all his sites (i.e. VLAN101) and that can be configured on his router to send/receive multicast streams completely independent from his PPPoE internet service
So what do I need in our network to make this work?
- The thing i’m concerned about is I want to isolate his multicast traffic. I don’t mind giving him a few unique multicast addresses, but I don’t want for instance to accidently leak OSPF multicast messages to his site from our routers. So I want some kind of filtering on there
- I imagine I need to install the multicast package on all routers that are in the path between his sites? OR can I instead get away with just multicast package at the edge routers and then using VPLS or some other tunnel? this would reduce any downtime in having to reboot other routers in our network to install the package
- PIM or IGMP? I imagine I need PIM in this instance, but every single multicast guide i’ve seen has both enabled. Is there any harm in having IGMP ticked?
(I noticed when using my test client (Multicast Test Tool) it wasn’t working when IGMP was unticked. Bug? Or user error?) - What are the actual steps here for this setup. In my mind I think I just need to define a RP, and that would be closest to the sender in this case. And then add every interface that faces towards the sender, and to the receivers, and only tick PIM? (or leave both on)
- In order to restrict/isolate the traffic do I just change the default group of 224.0.0.0/4 to i.e. 239.1.2.0/24 and that way he can only use 239.1.2.0-255 and it’ll never include any other multicast addresses? Or do I need to add some firewall rules?