Managed is preferred. Data packets are unicast (
point to
point) or multicast (
point to many). Most AV and lighting protocols use multicast because one source is sending the same data to many endpoints. There are other features too like QoS quality of service that prioritize expedient delivery of mission critical data over general internet traffic.
Unmanaged switches do not know how to handle multicast, so they take those packets and repeat them out to every port as broadcast (
point to everyone), which can big down and hose your entire
network.
Story goes at a Vegas casino someone accidentally patched the Cobranet
network on an unmanaged
switch into the
house network. The multicast traffic became broadcast and the ensuing broadcast storm crippled the gaming, hotel WiFi,
point of sale, and security systems.
In a small scale, isolated
network, it’s unlikely you would see any issues, but as a matter of form you should treat your
system backbone as a mission critical device. The unmanaged thing you buy for $30 may work well enough for now but could give you surprises when you least want them, and are generally more prone to individual port failures and
power supply failures. One bad port can also take down an entire
network.
Dante and
ETC both recommend Cisco SG300’s, which have recently been transitioned over to the SG350
line. This is what you will see the most of out in the wild. If you don’t have much
network management background, a Google adventure will yield step by step guides that are pretty digestible. Configuration shouldn’t take more than 30 minutes.
I don’t recommend specialty switches unless
etherCon is a must. Usually if
etherCon is a must, I stick an SG350 in a rack with a custom panel of etherCons and patch the
switch to the panel internally of the rack.
In general for specialty “Entertainment” switches, you’ll pay 3-5x the price per port that you would otherwise. You cannot Amazon Prime next day ship a replacement. And you are at the mercy of a
network R&D team who may understand what Dante,
sACN,
ArtNet, and AVB are, but have 1/10,000th of the R&D and development resources Cisco, Extreme Networks, or HP have. In general, it is an extra expense for which the average customer receives no tangible reward.
Ever want to get a good idea of how the R&D process isn’t the same for specialty switches, look at the firmware
release notes for the
Yamaha etherCon switches. They spent the first year or two of their product on the market publicly
bug testing obviously egregious flaws in their firmware using your show as a guinea pig. First time I used them was the last time I’ll ever use them.
Similar story with Pakedge. Their sales
pitch is that they make web management interfaces stupid simple for AV, and yes it was nice being able to
call someone in support who understood Dante and Q-LAN, but for three times the price we had a few switches in installs across the country going tits up with hardware failures or unresolved
network config issues that disappeared swapping out for Cisco.
There was another product I used a while back by Link, that was also 3x the price for them to take a Cisco SG300
PCB and stick it in their own chassis with lights and ethetcons on the
face panel, load their own settings into, and get the UL listing for. You basically are paying for them to be a middle man — and again, there are times where a repackaged product like that is valuable but 97% of people will never benefit from that in any way beyond what they would get from a quality off the shelf product.
Want a 48 port specialty
switch? Good luck finding one, and if you mix and match the 8-12-16 port specialty switches with a Cisco 48 port
switch at the core of your
system, good luck getting either manufacturer to help you troubleshoot why your multicast is falling apart at the transition between them.
By the way “Entertainment Class” mentioned above isn’t a thing. It describes a market those switches are intended for but does not exist as any industry standard or otherwise formal
category of products or protocols.
Also,
ethercon isn’t all its cracked up to be. It’s expensive, there are 3 versions in the wild right now each with its own quirks and incompatibilities, and the general wisdom used to be that you could always unscrew the shells on either end of you needed to shove an RJ45 into a non
ethercon jack. Guess what? There are versions out there now where the pins are built out of a
PCB board that fall out when you pull the shell off. Now if you want to use
ethercon in a meaningful way, better have an inventory of ethercon-to-RJ45 in addition to your ethercon-to-ethercon cables.
None of this is to say that specialty switches or
ethercon connectors have no useful applications, but as a design consultant I have gotten past the fad of using these niche products just because they’re purpose-built. Their cost eats into other goals I need to achieve and they add a sometimes unwelcome additional layer of complexity to projects, regardless of whatever “Oh, just set the
dip switch for Dante and it’ll automatically configure everything for you” claims they may make.