Stacking vs. LAG ports for AV over IP

This article was authored by Michael Dijk, System Engineer, Pro AV at Netgear

Stacking in general is very good, because it allows one management address, and avoids you having to login to numerous IP addresses for management.

However, in the AV world, where Multicast is used, it’s inefficient for Multicast traffic, the reason for this is that Multicast is flooded across all stack members through the stack links.

Therefore, when stack links aren’t correctly built, or calculated, there isn’t enough bandwidth for the network to operate correctly. Thus, saturating the links, and causing traffic to drop.

 Our advice, when designing networks, is to only stack the core, with sufficient bandwidth between the stack, to allow for this. 

Therefore, you must use LAGs instead, to have the bandwidth available, the advantages of LAGs:

  • Scalability, there is no hard limit on how many switches can be connected.
  • Efficient handling of multicast traffic With IGMP+, only the required multicast traffic transits between switches.

 For example, if you had a stack of 5 units, and these units where all daisy-chained together. They would only utilize one link (10gbit) between from one to another. If more than 10 streams would have to traverse from switch five to switch one, that would saturate that link, because we need to consider the traffic for video + stack traffic/data + other data. 

With LAGs, to the core, this won’t be an issue, because we calculated four ports per switch using 10gbit, giving you bandwidth of 40gbit, thus sufficient to handle all the streams that do need to route across the network.