I need some advice on best practices when working with multicast. We are upgrading our audio routing distribution system and it is centered around multicast. I work at a radio network and we can't afford more than a 30 second outage. The upgrade is going to introduce 3 new switches (for 3 studios) into our network (possibly meraki or catalyst, not sure which yet) but I would like some advice on what people have found to make the multicast traffic highly available. Should I implement RSTP? is there a preferred topology? Is there an edge to using meraki switches over catalyst 2960's? I found a page https://www.cisco.com/c/en/us/products/collateral/ios-nx-os-software/multicast-enterprise/whitepaper... that shed a little light on this but wanted to reach out to the community.
I think you should be aiming high and considering something like a Cisco Catalyst 9404R with dual supervisors.
This solution has much higher levels of availalability, and you can do so many things without causing any downtime at all. This includes being able to do software upgrades with zero downtime. It will also have considerable multicast throughput. In general, they are amazing switches.
My second choice would be a stacked pair of Cisco Meraki MS390 switches. Apart from firmware upgrades you should be able to do most things with only seconds of downtime.
I probably would not go any lower end than a stacked pair of MS250 switches (make sure you get the redundant power supplies as well).
It is very important you keep your layer 2 architecture loop free. Use stacked switches. Use LACP for redundancy. Use RSTP for safety against human mistakes but don't rely on it for handling failover.
Just to make sure you got the most important bit - design a loop free layer 2 architecture.
Thanks for the good advice. I'll keep this in mind. Pretty much my whole audio transmission route is centered around multicast. If I have a layer 2 loop, why is that such a bad thing? The article I read explained it a little bit stating that the time of convergence was pretty latent with RSTP and said to use PVST or PVST+ if possible. Should I have no RSTP on this period? Doesn't the Meraki ports use RSTP port fast as default? How does LACP resolve a failover? I know a little bit about Link Aggregation but haven't actually needed to use it yet in my career.
One point I would bear in mind is that however you implement the solution you eventually deploy, you should avoid routing multicast through a Meraki MX security appliance/router, as, in general, a number of multicast services such as an IGMP-Proxy, have not yet been implemented.
Meraki switches are more multicast capable, but I would still trial a couple of their switches prior to committing to any specific hardware. I have found, with multicast (video and audio) that actual implementations of multicast services are variable and that end devices can have quite specific requirements which are not always immediately apparent.
I prefer to isolate multicast traffic from other network activity by extensive use of VLANs, avoidance of native LANs and implementing explicit, generally non permissive, switchport profiles on all ports. A little fiddly to set up but it pays off every day thereafter.
I'm sorry @Uberseehandel, but your thumbnail and general demeanor on the forums....I can't resist. 😋
At least you didn't describe me as a Guild Navigator
Thank you for the advice. Like you said, no IGMP proxy for the MX security appliances. Should I be utilizing the L3 portion of the MS225's? If I wanted to prepare for the future and say, send multicast between our stations through a site-to-site VPN, how should I prepare for this architecturally?
>If I have a layer 2 loop, why is that such a bad thing?
Because a loop free design is the most stable design, and uptime seems important to you.
Also note Meraki VPN can not transport Multicast traffic. You'll have to use traditional Cisco routers and use GRE over IPSec (use a VTI).
Are your multicast sources on the same subnet/VLAN as the multicast receivers?
If they are on different VLAN's where does the interVLAN routing happen? On the CORE switches or Firewall? Or do you have multiple hops between your multicast sources and receivers?
If they are on the same VLAN. Are you running PIM on the first hop or do you use IGMP querier on the CORE switch?