I’ve been in the process of gathering NVX to replace DM in my home system and I’m to the tipping point where I’m one FedEx delivery away from having “enough” that I can pull out DM and survive without a few sources and a few destinations.

?

Currently the “AV Closet” has a couple fairly old Cisco 12-port 10/100/1000 fanless switches uplinked to my core. Obviously that won’t support NVX well. The existing traffic on the switches is pretty light – processor <> processor, processor <> touchpanel (via core), Chromecast, Apple TV, etc.

?

The AV Closet will (eventually) have 2 DMF-CI8s and possibly a loose decoder or two – worst case though is 16 encode and at 750MB that’s about 12Gb/s (or 16 Gb/s if we assume full line rate from every encoder). For the most part I’m planning on hanging the decoders in rooms off the existing DM CAT however there are a few places where feeding them from the core makes more sense (and currently I’m doing weird DM cross-connects to make it work).

?

My current plan is to replace both of the current switches with a Cisco 3850 12X48U (12x10G ports and 36 1G ports) and to LAG some number of the 10G ports back to the core.

?

My internal debate that I’m curious about the collective thoughts:

?

1. Cisco’s implementation of LAG is only optimal with an even pair of links in the group. 2x10G would give me 20G theoretical with 4G of headroom which seems like it should be plenty considering everything else is running in a 2G aggregate link and not even hitting 1% utilization… but should I just double than to 4x10G (40G aggregate) since I have the cable path and ports available and I’m suspicious that now that “DM jacks” and “LAN jacks” will be interchangeable traffic from things in the field may wind up hitting this switch on the way to the core?

?

2. What are the feelings about throw NVX on my existing “Crestron/IoT” VLAN (with the existing IGMP querier on the Core) or create a new AVoIP VLAN with the querier running on the “AV” switch? I’m torn because keeping it on the core seems like the most “by the book” answer and with 20 or 40g between the switches shouldn’t have any traffic/blocking concerns but on the other hand if virtually all of the encoders live on the AV switch it seems kind of stupid to forward that traffic to the core just for the sake of having all of the IGMP queriers live in the same place. ?(I’m also considering playing with Layer 3 multicast routing which is heretofore something I’ve never even dipped my toe in the water on)

?

Thoughts? I don’t think there is a single “right answer” to either question but am curious about thoughts and approaches from the community

?

?

Lincoln

?

?

?

--

Lincoln King-Cliby

Commercial Market Director
Sr. Systems Architect | Crestron Certified Master Programmer (Diamond)
ControlWorks Consulting, LLC
Direct: (+1)440.771.4807 | Cleveland: (+1)440.449.1100? | Boston: (+1)508.695.0188 | DC: (+1)202.381.9070? | London: (+44) (0)20 4520 4600
Crestron Services Provider | Biamp Authorized Independent Programmers | Extron Qualified Independent Programmer

?

There are about 7 decode (plus 1-3 encode, currently I have 3 DM-TXes around the house but only really use one with any kind of regularity and the other two are definitely in the “they’ll wait until I get my hands on more gear”) and the distribution of encoders and decoders is largely to get out of the (IMO) stupid cross patching I’m currently doing to get DM to and from data drops in places where “I’ll never need AV CAT cable here…” (from the time before DM was really a thing… but remind me never to say that about LV cabling next time I buy and wire a house…)

?

The switch choice aside from being in my “reasonable enough” price range (it was about $80 shipped including both power supplies) was mostly driven by the desire to be able to do 10G over copper between the “Data Core” and ?“AV Core” but also because I wanted to be able to power the non-“C”-series cards over PoE – and the DM-NVX-350 are somewhat absurdly power hungry (requiring the 60-watt UPoE capability) ?(Aside from just not wanting to deal with wall warts, I manage power somewhat aggressively on the house when I’m away so being able to turn off PoE to the switch port is a lot cleaner than scattering a bunch of relays or controlled outlets around.

?

Since I have the cable between AV and data and the 10G ports are already there the fat pipe is essentially free.

?

Looking forward to having some fun – all the weird things I’ve been curious about but would never do with a client’s hardware or on a client’s network….

?

Lincoln

?

Funny coincidence – I?m currently at the exact same point: replacing my the current DM-stuff in my house with NVX.

The DMF-chassis with 6x350C arrived last week and is already up and running and I?ve collected a bunch of 350 and D30/E30 over time that are ready to go.

I already have a fairly reasonable network in place all based on Netgear M4300 and M42500 switches, interconnected via 10G fiber.

?

My setup won?t be as large as your?s. I?m looking only at 6x350C installed in the chassis that I plan to use as encoders and just about 3-4 projectors and displays for the decoders. All wired up nicely with CAT6 when we moved in 10 years ago. ?

?

With regards to bandwidth: I?d plan the bandwidth requirements on actual use, not worst-case. At least for my own house. So which/how many streams are you actually about to use at the same time? How likely is it that all your encoders and decoders will transmit/receive streams at the same time?
I think you should be able to scale down the uplinks.

?

With regards to the VLAN/querier question: querier data is just metadata and not the actual streams. So I would go “by the book”. At least that?s what I?ll do. I?ve put all the NVX stuff on a dedicated VLAN with the querier remaining on the core.

?

Cheers,

Thorsten

?

?

There are about 7 decode (plus 1-3 encode, currently I have 3 DM-TXes around the house but only really use one with any kind of regularity and the other two are definitely in the “they’ll wait until I get my hands on more gear”) and the distribution of encoders and decoders is largely to get out of the (IMO) stupid cross patching I’m currently doing to get DM to and from data drops in places where “I’ll never need AV CAT cable here…” (from the time before DM was really a thing… but remind me never to say that about LV cabling next time I buy and wire a house…)

?

The switch choice aside from being in my “reasonable enough” price range (it was about $80 shipped including both power supplies) was mostly driven by the desire to be able to do 10G over copper between the “Data Core” and ?“AV Core” but also because I wanted to be able to power the non-“C”-series cards over PoE – and the DM-NVX-350 are somewhat absurdly power hungry (requiring the 60-watt UPoE capability) ?(Aside from just not wanting to deal with wall warts, I manage power somewhat aggressively on the house when I’m away so being able to turn off PoE to the switch port is a lot cleaner than scattering a bunch of relays or controlled outlets around.

?

Since I have the cable between AV and data and the 10G ports are already there the fat pipe is essentially free.

?

Looking forward to having some fun – all the weird things I’ve been curious about but would never do with a client’s hardware or on a client’s network….

?

Lincoln

?