Jump to content

DMX Over Ethernet and Fibre Optic


goochr96

Recommended Posts

If you are going to use anything that uses ArtNet, on a shared campus-wide network, you need to be very careful. Artnet allocates its own IP addresses, based on MAC addresses and uses addresses that have special meanings and should not be used by individual hosts according to the internet protocol specifications (e.g. x.x.x.0 - address of network and x.x.x.255 - network broadcast address). At best, network routers will drop all packets to these addresses!

 

ArtNet also conflicts with MSN Messenger.

 

If I were trying to implement such a widespread system, I would establish a VPN (virtual private network) tunnelled between buildings over your campus network, isolating the ArtNet data from the main network.

 

Something like the NetGear RP614 can be configured to establish a VPN between several units - your campus network experts will have their preferred solutions. If they get a chance to read the ArtNet specification and understand the implications, they are likely to insist that you do it this way (if they let you use their network at all!)

Link to comment
Share on other sites

  • Replies 32
  • Created
  • Last Reply

I've never had any dealings with ArtNet, but (wearing my network engineer's hat) I've just skim-read the spec.

 

My first reaction is..... Yuk!

 

The addressing is a bit bonkers. It works, of course, but makes huge assumptions about the architecture your network. Basically, it assumes that there aren't any other devices, and that there is layer 2 connectivity (ie bridging, not routing) between all nodes. Rather strange. It's a bit like making a rule that we'll drive on the right and give way to traffic coming on to a roundabout, knowing full well that everyone else on the road does precisely the opposite....

 

I know that if one of these devices was plugged onto my production network, the port would be automatically shut down, and the network police would be knocking on the door....

 

But it will work perfectly well on an isolated network.

 

So, it basically needs an isolated "island" overlaid on your production network. VPN might be an option, but knowing the way University networks are normally designed, just implementing a separate unrouted vlan would be more appropriate and efficient, and assuming you've got a reasonably modern infrastrcuture, much cheaper.

 

If anyone plugged a NetGear RP614 or similar device into my production network without explicit permission, they'd get (a) immediately chopped off, (b) a visit from the Network Police, and © a 300 pound fine. And I can't see me giving that permission :). The potential for disruption to other users of the network is immense!

 

Bruce.

Link to comment
Share on other sites

Even though we support ArtNet, I'd have to agree that the standard leaves a bit to be desired. It is, undoubtedly, a poor fit for an existing LAN/WAN. The packets are broadcast UDP. The obviously very network savvy group here can probably best advise on rather to look at a tunnelling protocol like VPN or some other solution.

 

We've been used in quite a few theme park and architectural lighting applicatons and generally programming is handled one of two ways:

 

Single Controller - one controller, distributed DMX

Syncronized Controllers - multiple controllers, typically synced via SMPTE

 

The former is actually one of the reasons that we support multiple, concurrent timelines for timecoded shows. In arch or themepark, a given audience member may not 'see' the same vantage point as the programmer, so you want to keep things interesting at each viable locale. Concurrent timelines lets you 'layer' the show instead of collapsing all elements into a single sequence of scenes (like you would in timecoding a show on, say, a Hog II).

 

Synchronizing controllers is actually, I think, a very sensible way to go. When you examine the costs of distributing DMX over a large area, placing a controller at each locale can potentially be a cost savings (at least for us). SMPTE represents a miniscule amount of LAN traffic compared to a single universe of ArtNet DMX and, being a relatively low frequency audio signal, is easy to transmit wirelessly as well. Programming may seem less intuitive, but distributed systems also represent a form of 'layering', giving the ability to program each locale without having any concerns about other scenes and fades. Multiple systems also generally means that you are less likely to lose an entire show do to a single piece of equipment failing. Most systems that accept timecode like us also have a concept of dealing with lost timecode - switching automatically to 'internal' timecode with the external system is lost.

 

-jjf

Link to comment
Share on other sites

I've never had any dealings with ArtNet, but (wearing my network engineer's hat) I've just skim-read the spec.

 

My first reaction is..... Yuk!

 

The addressing is a bit bonkers. It works, of course, but makes huge assumptions about the architecture your network. Basically, it assumes that there aren't any other devices, and that there is layer 2 connectivity (ie bridging, not routing) between all nodes. Rather strange. It's a bit like making a rule that we'll drive on the right and give way to traffic coming on to a roundabout, knowing full well that everyone else on the road does precisely the opposite....

 

On the face of it, in default mode it is horrible however it dosn't have to be that bad.

 

Okay, so it uses UDP/IP so it'll drop packets however it can be assigned an address and work in unicast in what it calls "server to peer" also being connectionless its faster, I'm sure the gateway abstract lost packets.

 

In short, it'll work and you shouldn't need a VPN or a trunked VLAN, however you'll have to configure it away from the network before you connect it up.

 

http://www.artisticlicence.com/user-guides/art-net.pdf

Link to comment
Share on other sites

Okay, so it uses UDP/IP so it'll drop packets however it can be assigned an address and work in unicast in what it calls "server to peer" also being connectionless its faster, I'm sure the gateway abstract lost packets.

 

It should be noted that quite a few vendor's ArtNet support does not handle non broadcast packets correctly. Also, I believe that AL changed the meaning of the 'only to me' bit. I've yet to find any gear that properly handles packets out of sequence (the largely unused sequence field). A lot depends on the sockets implementation used under the software. For example, I've noticed that LanBox sends every UDP broadcast packet 3 times. I doubt that this was by design, but a 'feature' of the socket implementation.

 

I would expect the university network police to get angry if you hooked us up and started sending out 16 full universes of packets 30 times a second. I can only imagine if everything was being sent 3 times!

 

-jjf

Link to comment
Share on other sites

I would expect the university network police to get angry if you hooked us up and started sending out 16 full universes of packets 30 times a second. I can only imagine if everything was being sent 3 times!

I used to (groping for words...) look after networks for several City of London institutions for most of the 90s, and yes, I could get the handbags out at dawn with the best of them.

 

If this arrived as a "must support" application the least 'orrible approach (assuming you've got expensive hardware, which I would expect a university so to have) looks like using VLAN trunking to provide artnet with its own little flat network within the existing infrastructure. VLANs possible but hard and ugly when what you really want is flat. If the controller needs to be on the "real" LAN use a router to NAT to the artnet LAN in its address space. ugh.

 

When I heard about artnet a while ago I got all excited till I read the spec and haven't looked since. It was clear that it presented certain "real world" constraints <_<

Link to comment
Share on other sites

When I heard about artnet a while ago I got all excited till I read the spec and haven't looked since. It was clear that it presented certain "real world" constraints <_<

 

Hey but look at the bright side, as long as converagance continues to be hard work, there is money to be made!

Link to comment
Share on other sites

Just a thought, but why is it when someone wants to light up the centre of Coventry everyone wants to tell that person how it won't work, how the spec is wrong and how the network will meltdown immediately if anything other than the VCs request for catering is sent on it.

Compare and contrast the attitude for lighting up large bits of Hong Kong harbour using various bits of the public communications system - and, indeed, various individual commercial networks:

 

http://www.laservision.com.au/hongkong/

 

A mere observation. ArtNet, for all its faults, is at least an attempt to solve the issues of lighting data over TCP/IP and UDP. If we were to all sit around and wait for ACN, and I admit there are some very complicated issues about Device Description Language and Device Management Protocol, then the earth will have cooled. It's open source, so if we don't like it we can change it.

 

Cheers

 

Ken

Link to comment
Share on other sites

Just a thought, but why is it when someone wants to light up the centre of Coventry everyone wants to tell that person how it won't work, how the spec is wrong and how the network will meltdown immediately if anything other than the VCs request for catering is sent on it.

 

I don't think anyone said that. I certainly didn't, and I'd guess I'm one of the most qualified to comment, as I design/manage/fix/break a very similar University network for a living.... If I configured OUR campus network to support this, it would cope without breaking sweat. Compared to some of the real-time services we already offer (telephony, video streaming, CCTV, environmental monitoring and building management etc), the bandwidth and latency requirements are fairly trivial.

 

The point is that no wide-area production network will be configured in the necessary way by default. Some changes will need to be made to provide a wide-area layer-2 broadcast domain. Not difficult, just not usual!

 

So get the networking people on board very early in the process. Knowing the sort of people who work in this field in Universities, I'd expect them to be keen to get involved. But if you just turn up unannounced, having spent a pile of money on a pile of boxes, and say "we need this working by tomorrow morning", the response may not be as helpful. Believe me, I'm speaking from experience!

 

Bruce.

Link to comment
Share on other sites

Just a thought, but why is it when someone wants to light up the centre of Coventry everyone wants to tell that person how it won't work, how the spec is wrong and how the network will meltdown immediately if anything other than the VCs request for catering is sent on it.

<snip>

A mere observation. ArtNet, for all its faults, is at least an attempt to solve the issues of lighting data over TCP/IP and UDP. If we were to all sit around and wait for ACN, and I admit there are some very complicated issues about Device Description Language and Device Management Protocol, then the earth will have cooled. It's open source, so if we don't like it we can change it.

 

The central messages we are giving is

 

*It's perfectly possible if you get a good network guy for the university on-side

*You'll need to configure the devices before attaching them to the network

 

I personally am all for convergence and would like to do it full time and I think the efforts of artistic licence, ethersound, cobranet, stardraw-control et al is too be applauded.

Link to comment
Share on other sites

Okay then, so I think we are all agreed that although the AL DMX over LAN is a good device, when installed in an isolated network. It does however have quite a few "serious" draw backs (as have been kindly pointed out), so may not be the choice of device for this application. Do we know wether the LAN Box has the faults, also would Strand's networking device work more happily along side our networked services? But dont worry, I am not just going to go out and order one, I will run things past the networking guys!!
Link to comment
Share on other sites

Hi there all,

I showed the manual for the AL DMX to LAN to a friend of mine who works is an IT bod for the RSC in Stratford, and he just shuddered and shouted "Filth!!" He did say that if AL were to change the firmware on it, it would be a really nice deivce. Also showed him the Strand node, and he agreed that it looked a lot more sensible, with a sensible IP address! Apparently though, Strand have told them that if they install a network in the RSC, to put nothing else on it apart from there (Strand) stuff, but I think this is only because it is having a Pc on it that has no anti virus software on it.

Link to comment
Share on other sites

How about a nice small mac mini at each location with one of my USB DMX Interfaces on it?

 

If you have the specs / ideas on what you want your building project to do I'd be happy to work with you on it to write the relevant software needed for the minis etc.

 

If your interested email me, or drop me a PM and we can discuss it further.

 

Ben

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.