JANDS NEWS

TECH CORNER: Combining Your Ethernet Networks: A Good Idea?
06 September 2010

Ethernet. You may have heard of it: apparently it’s going to change the way our stuff talks to our other stuff. In the audio world, solutions such as Cobra-Net use Ethernet to cheaply and reliably send digital audio data between devices. In the lighting world, ArtNet and Streaming ACN are used for sending DMX data to fixtures. And then there’s all of the monitoring, configuration and backup systems that also make use of Ethernet for communication in both audio and lighting equipment. High-definition video will probably go this way too, and the vast majority of general purpose computer networks have been running on Ethernet for many years now.

So, since everyone’s moving to this common transport protocol, why not combine them together on one super network? On the surface, there are some obvious advantages: less Ethernet infrastructure required, only one network to manage, and you can integrate any common functions between the disciplines. But, when you look at such a solution in a real-world situation, it gets a bit more complicated. Not only are there bandwidth considerations, some audio protocols are designed to sit alone on networks. There’s also the human factors: who’s going to take responsibility for any faults or issues on the network? How can you ensure the other departments don’t clog up the system with non-critical data? Well, that’s what we’re going to talk about today. Hopefully, by the end of it, you’ll have a clear idea as to whether combining the Ethernet networks in your new or existing installation is a good idea.

The Advantages

I’m going to be upfront: I think that in many situations in our industry, combining all of your Ethernet networks is a bad idea. But before we examine the reasons why, let’s make sure we’ve examined the advantages of combining the system first.

Infrastructure Costs

An often quoted advantage is that you can reduce the quantity of CAT 5 and Ethernet switches required in a system. In reality, though, you still need multiple CAT5 runs to each location, to cater for multiple devices in that location. So, in reality, a heavily used location may differ from 6 points all on one network to 8 points on four separate networks. Given how inexpensive installation CAT 5 is, there’s actually very little in cost savings. It’s a similar story with switches, going from one big network to four separate networks generally requires only a modest increase in the quantity of extra switches required. While a good switch isn’t as cheap as extra CAT 5, it’s still pretty inexpensive when compared against all the other specialist audio and lighting equipment going in to a job. All up, it seems as if reduced infrastructure costs is often a false economy.

Sharing Data Between Disciplines

The other big advantage is for cross-communication between disciplines. Many lighting controllers synchronise with an audio track via MIDI or SMPTE timecode, which could well be coming from something on your audio network. Similarly, you’ll want to ensure the video data running on the video network is synchronised to the audio data running on the audio network. If they were on the same network, they could sync their genlock or audio clock without any extra mucking around. In a venue, sharing your system with the office network could allow you to do basic monitoring or configuration from your desk in the office, rather than needing to walk down to the exhibition room or theatre. All of these benefits rely on data being shared between audio, video, lighting and office networks. To separate them out removes these benefits in the worst case, and requires extra linking and mucking around in the best case.

The Disadvantages

Capacity

Video (particularly in high-definition) is very bandwidth intensive. Likewise, many channels of uncompressed audio at high sample rates can dominate the available bandwidth of a system. It’s important to keep the used bandwidth in current networks comfortably below the practical maximum to avoid high latencies. A fully-loaded or mostly-loaded network will have to stop and delay some traffic to let others through: similar to how road traffic tends to queue up in peak hour. As a result, noticeable delays or lags can be introduced on to the network: a disaster for audio, and pretty unpleasant for lighting and video.

A related issue has to do with the low-level format of protocols such as Cobra-Net which only use the low-level components of Ethernet (more correctly, they implement Ethernet and not the higher-level TCP/IP protocols which are part of most other Ethernet traffic). Cobra-Net has its own parts added on to the Ethernet protocol which assist in sync-ing packets and keeping the latency low. All of this works just fine on otherwise silent networks. On networks with a significant amount of non-Cobra-Net traffic, however, all bets are off. It’s not just Cobra-Net, either. There’s a few audio and video systems out there that do similar things, and these systems are best left on their own private network.

Ethernet technologies such as Quality of Service (QoS) can be utilised to reduce both of these considerations. However, QoS is tricky to both configure and verify accurately, and can be quite limited in its configuration options on many switches. Also, it merely prioritises some traffic over others, so the lower-priority traffic will experience even slower speeds and longer latencies than before. Without a qualified IT technician to program and maintain all the traffic rules to keep everything running smoothly, separate networks with plenty of extra bandwidth and clearly defined duties is a much simpler way of keeping these issues in check.

Responsibility

When evaluating these issues, it’s important to consider not just the technical feasibility of the system, but the maintainability of the system. Technology can and will fail at some point, and at that point someone has to be able to on hand to quickly find and repair the fault. It’s all very well combining all the audio, video and lighting equipment on the one network, but if something goes wrong (or is suspected to have gone wrong), is it the audio, video or lighting guy’s job to fix it? If there’s a clear IT-type person on the show, then this is less of an issue, but many shows or installations do not have such a person on staff. In separate networks, not only is less IT expertise required to keep the system going, but it’s clear who’s responsible for which network: the audio guys are responsible for the audio network, the lighting guys for the lighting network and so on.

In a similar vein, it’s important to have trust within the network. In small, self-contained networks, it’s well understood what the network is used for. A lighting network will only get used for lighting purposes, while the office network will only get used for that. As soon as these various networks start to combine, boundaries get pushed. No matter what goes in the policy handbook, it’s inevitable that someone’s going to try playing a network game, or transfer a 100MB file between computers and disrupt the communication of other, more critical traffic. If during a show or presentation, this could have a disastrous affect on the audio or lighting. It may seem that these sorts of issues can be controlled with a few stern words or signs, but would staff in a different department (the director or client, for example) really appreciate the effect their seemingly basic computer operations can have on the integrity of the audio, video or lighting?

Looking to The Future

As both the entertainment protocols and the underlying Ethernet protocols are refined, the disadvantages in combining networks are slowly being resolved one by one. Probably the most exciting development is Audio Video Bridge (AVB) which is a few extensions to the IEEE802.1 suite of Ethernet protocols. This will provide a number of standard, IEEE-ratified mechanisms for synchronising and reserving streams which will keep the latency predictably low, no matter what other traffic on the network exists. AVB is being designed with the automotive, consumer and professional AV markets in mind, and a number of professional audio manufacturers have already begun showing interest in it. Harman has been investing heavily in AVB and has already begun showing AVB-enabled products at industry trade shows.

So, is it Worth It?

Well, that’s a decision that depends on the requirements and running of the venue or installation. Systems that are frequently reconfigured, have many untrained users configuring or using it, or work in show-critical environments probably have more to lose than to gain from combining all the networks together.

Larger networks with dedicated and prompt IT support, set-and-forget installations or smaller environments where audio, lighting and video departments are all the same person may find the disadvantages of combining the network less of an issue, and are happy to work with one larger network. Even in these cases, keeping the office network separate (at least via VLAN, if not physically) is likely to keep things simpler and more reliable, and the issues regarding the interoperability of low-level protocols such as Cobra-Net with higher-level protocols such as Streaming ACN still need to be considered and tested.

If technologies such as AVB grow in prominence in the coming years, we could well see the disadvantages of combining networks diminish significantly, such that the majority of installations can happily combine their networks without issue. Until then, however, designing a system or installation around multiple Ethernet networks is likely to be a safer and surer way of keeping everything working reliably. Simple colour coding of the RJ45 or Ethercon connectors can show which ports in a venue correspond to which network, and the increase in the cost to the project of extra CAT 5, connectors and switches is usually pretty marginal. And, if combining networks becomes a convenient and viable solution in the future, then it’s as simple as one link lead between each network to combine them.

Article courtesy of Jands Australia, jands.com.au