AVB, Audinate Dante, and AES-67: Peering into My Crystal Ball
Update June 25 2018: Latest thoughts here
June 24, 2014: Note, an update to this entry is posted here.
July 2, 2015: Note an even-more current update to this blog entry is posted here. I would suggest reading it first.
Update June 25, 2016: I have yet another followup here.
Several events before and during the AES convention last month here in NYC indicated for me a future for the world of audio networking for the live sound industry. Before I reveal what I saw in my crystal ball, let’s review some recent developments.
AVB
I first really became aware of the IEEE 802.1 standards effort for Audio-Video Bridging (AVB) back at a demo on the floor of the Infocomm show in Vegas in 2009 (my writeup and photos here). AVB’s goal was to create an open, free to use network for low-latency, high quality audio and video distribution natively over Ethernet. Seeing working devices on the Infocomm floor in 2009, I thought this cool new technology might come on the market soon afterwards. I was wrong. Instead, AVB languished through the glacial IEEE standards-making process; every time I asked, everyone I knew who was involved said “oh we’re just waiting on this one last thing" (that one last, final thing just came out a couple months ago). At Infocomm in June of this year, there were a few encouraging signs (my writeup here) that we might finally, officially get AVB out into the world in real product. But then in July, a month after Infocomm, on my way back from a photo workshop in Maine, I stopped in to check out the AVB testing program sponsored by the AVNU Alliance (the trade association created to promote AVB) at the Interoperability Lab (IOL) of the University of New Hampshire. It’s an impressive lab (my writeup and some photos here), but I’m afraid AVB might be too late to the live sound party (and I voiced this concern to the lab director during my visit). In July, they were—more than four years after I first saw working prototype AVB equipment at Infocomm 2009 in Vegas—just testing new, AVB-capable Ethernet switches to join the few already on the market. The problem this causes is that AVB requires those special AVB-capable switches to function properly, so without a wide array of these on the market the standard is effectively stalled out from the user's perspective. And worse, the rumor I heard from several people at AES was that Cisco--the 800 pound gorilla of the networking world--was not going to implement AVB in their switches any time soon, despite being a founding member of the AVNU alliance.
In the Meantime: Dante
During these four years, of course, the market did not stand still; Audinate’s proprietary Dante audio network solution basically took over the live sound market. The turning point for me was when Yamaha announced at Infocomm in 2012 that their CL series live sound consoles would be based completely around Dante. And around the time of my IOL visit, an article came out from Audinate talking about the situation (linked from my writeup here), which basically said that while they supported AVB, they—and the pro sound market—were not able to wait forever for affordable AVB-capable switches. They also speculated that AVB switches didn't offer compelling value to our market, at least in the near term with high costs and limited availability. The only viable possible path forward they saw was that AVB “…may gradually diffuse throughout the industry and become available in a wide variety of network equipment.” As I wrote then, “After watching this market for 25 years, I've found it rare for technologies like this to gradually 'diffuse' through the industry.”
I tweeted my July blog entry at Audinate, and US-based director of Audinate, Lee Ellison, contacted me, saying he would be in NYC for AES and would like to get together. I of course said yes and took advantage of this opportunity to ask that he talk to my sound engineering students, and he gave us a great presentation. He is a very smart guy who understands way more about the technology than I would expect of the director of a company, and he clearly gets the market and the role of standards in industry.
Mr. Ellison was clear in the talk and at lunch afterwards: Audinate plans to continue support for AVB. But, because Dante is both a proprietary transport technology and also a sort of “glue” that can tie together disparate transport standards, it seems to me that Audinate will not use AVB as their primary backbone (there’s so few AVB capable Ethernet switches available today it’s hard to do in any case), but instead offer AVB as an option to those who want it. But Mr. Ellison also talked a lot about a new and potentially game changing standard: the then just ratified AES67-2013: AES standard for audio applications of networks - High-performance streaming audio-over-IP interoperability.
AES67
Back at an Infocomm AVNU alliance cocktail party in 2011, I ran into my old friend Kevin Gross, the primary inventor and developer of Cobranet, which until Dante came on the scene was the most widely used networking standard for transporting audio in the pro Audio market. I asked him if he was working on AVB, and he said he was instead working on a new Layer 3 audio transport protocol. I didn't actually grasp the full significance of this standards effort until I heard the details at this year's AES convention, since AES67 was ratified and released this past September. Kevin guided the process to completion in about two years, which is an incredibly speedy time frame for the standards world (but what the market needs).
To understand the significance of AES67, you have to understand a little of the ISO layer stack (and if you have questions, I know a good book you can buy that explains it all). To summarize, though, AVB--as part of Ethernet--operates at Layer 2 only. That means that special AVB-capable Ethernet switches can understand and transport it, but you can’t easily send AVB between networks (either local or over the internet), because today that’s done primarily using Layer 3, and the Internet Protocol (IP, as in IP addresses). Dante and several other network systems operate at Layer 3.
All this means that with AES67 we now have an open-standard, Layer 3, modern way of getting high quailty audio between networks (more technical details here). So if someone creates a connection between an AVB product or system and AES67, then AES67 can also be used to get audio over to Dante or another network. And this is great news for us today, since peering into my crystal ball, this is how I see the live sound networking world in the near future:
All manufacturers listed in the graphic have, to the best of my knowledge, working, actual product on the market using those networking technologies as of this writing. As you can see, the scales are tilted pretty far over towards the Dante side, with Meyer and a few other key live sound players on the AVB side. But AES67 can provide the bridge between those worlds for the users.
It’s All About the Users
One thing that is often lost in thinking about these kind of situations and technology development is that end users—who actually buy and use the stuff, after all—drive the market. And they are voting with their dollars. Sure, Audinate could switch over their entire Dante infrastructure to AVB, but why would they? If they did so now, everyone who has a current installation would have to buy new AVB switches, and these are not cheap--a few months ago, I priced out some Extreme Networks switches and the AVB option cost hundreds of dollars extra per switch, and what does that get me? Dante is doing everything I need right now on standard switch hardware, at layer 3. It gives me an affordable multi-channel, high quality, self-clocking, easy-to-use transport system with incredible patch flexibility (including any PC or Mac using the Virtual Sound Card). Here's a setup I did the other day here at City Tech (where I teach) to test our new JoeCo Dante-enabled recorders. It took me literally about two minutes to get this configured and running once I had it plugged up.
Is Dante a panacea? Of course not. It's still new and I still have issues with it. We had all kinds of mysterious problems on our last show (turned out to be mostly Windows firewall issues) and getting the firmware upgraded required several calls to Yamaha tech support (who were great). We've also had occasional word clock-style clicking/distortion problems, etc, and when those things happen it's typically not clear what the next troubleshooting step should be, so I hope Audinate will put more diagnostic capabilities into their control software. And for the future, I can easily foresee problems when you're connecting stuff from multiple manufacturers leading to finger pointing; but that would be the same case with AVB too (and possibly worse since there wouldn't be any common hardware).
Conclusion
I wrote back in June, “Will 2014 be the year of AVB, Dante, or AVB/Dante? ” After AES, I think the answer is pretty clear—Audinate's Dante has won the race, while AVB is still being tweaked by the mechanics over in the pit lane. Those who have committed to AVB (like Meyer—and I’m a big Meyer fan and user) should provide AES67 inter-operability for their systems so we can use it to connect to my Yamaha mixer, Shure wireless mics, JoeCo live recorder, Aviom monitor mixers, SMAART analyzer (via virtual sound card), etc etc etc—all of which are running Dante today.