<img height="1" width="1" src="https://www.facebook.com/tr?id=414634002484912&amp;ev=PageView &amp;noscript=1">
Donate
SMPTE stands with our friends, colleagues, and family in Ukraine, and with all people of Ukraine
Donate

The Nuances of Building IP-Based Systems

June 24, 2022

As discussed previously in Newswatch, the media industry’s transition to IP-based systems continues its rapid acceleration, but now, in Brad Gilmer’s opinion, that transition has entered a new phase. Gilmer, an industry consultant, SMPTE Fellow, and executive director for both the Video Services Forum (VSF) and the Advanced Media Workflow Association (AMWA), suggests this new phase involves a shift in focus from building IP-capable equipment to building IP-based systems.

“We’re at a point now where a lot of operations that used to be linear playout operations, satellite channels, or domestic over-the-air stations have all moved to the cloud, and in fact, the [Covid-19] pandemic really accelerated that change,” Gilmer explains. “The technology is now mostly ready and available, and economically, for large companies, it makes sense to [implement it] at this point, if they haven’t already. However, a lot of those large organizations are looking at building IP systems on premises, for the most part. And many organizations are also looking at how to integrate their on-premises equipment and IP facilities with things that are in the cloud—building systems that are a combination of those two things. SMPTE and other organizations have done a lot of work to develop IP standards and specifications that now make a lot of this possible, and vendors have spent a lot of time, effort, and money getting to where they have products that work in the IP space and have functionality that is similar to or in excess of what they had with SDI. 

“But it falls to customers and systems integrators hired by those customers to build full systems out of such products. And so, we’ve come to an interesting point in the IP transition. If you go back to the beginning of it, where was no standardized IP essence, no agreed-on way to exchange video and audio over IP. Then, SMPTE and the VSF put in a huge amount of effort, and eventually came up with the [Professional Media over Managed IP networks suite of standards known as ST-2110]. Similarly, AMWA put a lot of effort into NMOS [Networked Media Open Specifications] that explain how to make connections between senders and receivers in an IP domain.

“However, the next step is what I call the ‘loading dock problem.’ When I worked at [Turner Broadcasting], manufacturers would deliver equipment to the loading dock, but it was up to us to figure out how to integrate the equipment and build systems. In terms of the migration of the industry to IP, now is the time when we are moving from being early adopters of IP technology to a point where it is all becoming more mainstream, and that is a critical point. Now, it isn’t just about the 25 or 50 top media organizations in the world that have very deep technical and financial resources—it’s about everybody else in the marketplace that have to somehow implement these technologies. So, I believe we are at a critical point where, in order for users to be able to understand how the equipment works and how they need to build their systems to meet their business requirements, they need education more than anything else. We need to educate people in this industry about how to put IP systems together. We are now at a point where, if we are going to get wide adoption of IP technology, we have to increase the base of educated users who know how to do that.”

Gilmer emphasizes that the industry has begun this education process, pointing to events like the IP Showcase at NAB 2022 earlier his year, and reams of online material on the SMPTE site, a series of VSF produced videos available to all, and material from AIMS, the IEEE, and many other sources. But he also emphasizes that it can be complicated figuring out where one needs to start in order to develop a good base level of knowledge about the development and implementation of IP-based systems.

“It depends quite a bit on where the individual is coming from, as to where they need to start in this education,” he says. “We’ve all, by now, pretty much had some experience with computers and home routers and dealing with WIFI connections, and so on. So that is a base level of knowledge of networking. But then, some engineers I have talked to who are career SDI people have told me they really needed to take some IP basics courses that were not video specific to really understand better how IP technology works, such as how IP routing works, and so on. Having good knowledge about those things before you delve into specific network configurations that allow us to do very precise timing of video and networking, for example, can really make the transition easier.”

Gilmer adds that the education process for creating IP-based facilities and systems also requires comprehension of the many available building blocks for such systems, and what their current status is. He emphasizes that the Technology Pyramid for Media Nodes produced in recent years by the EBU, clearly illustrates the importance of understanding not only what is possible, but also what is currently feasible. As pointed out in Newswatch in 2021, a large chunk of that illustration remains colored in red, meaning certain layers of a typical IP-based media facility’s infrastructure have not developed to a point where they are standardized, ubiquitous, or easy to understand and implement by the industry at large.

The point, he adds, is that the holy trinity of IP-friendly operations—flexibility, interoperability, and plug-and-play capability—remain hotly pursued goals, sometimes feasible in particular situations and sometimes not, rather than widely available characteristics.

“When you view the Pyramid, you have to understand those colors are based on not just what SMPTE or any other particular standards’ body has done, but rather they are based on the EBU’s assessment of how ubiquitous technology that can do those things is,” he relates. “One criticism I hear about the Pyramid is that so much of it is still in the red section. But the thing to realize about that is not that the industry is failing to move toward providing the user requirements that have been outlined by the EBU. Rather, it’s because it simply takes significant time for the industry to reach agreement about all these matters, and then publish standards, and then have manufacturers adopt them and put them into ubiquitous products.

“If you want plug-and-play, you may have to give up some flexibility in terms of what you can do, for instance. So, it is correct that the entire ST 2110/NMOS ecosystem is trying to satisfy overall requirements for video and audio, but we still have a lot of work to do to bring everything seamlessly together.”

One example of the constantly shifting landscape that makes it difficult to quickly put all necessary elements together is the fact that a series of new, specialized data transport methods are starting to pop up for various applications, and as they do, they need to be folded into the larger equation. Gilmer points to the industry’s ongoing work figuring out how NMOS controller specifications might interface with new technologies that target the Pro AV market, for instance—protocols like Dante (a proprietary audio transport protocol from a company called Audinate);  NDI (Network Device Interface, a royalty-free approach designed to enable low-latency video for live production applications); and the proposed IPMX scheme (Internet Protocol Media Experience—a set of open specifications to enable audio, video, and data information over IP networks for Pro Av work), among others.

“AMWA has been looking into how NMOS controller specifications might work with these other transports,” he says. “I expect these sorts of technologies will co-exist with a ST 2110/NMOS environment in the long run. If you look at IPMX, for instance, the effort there is to use IP video in a Pro AV environment, and so, we expect there will be some aspects of the 2110 specification that could be simplified, or have some of its options reduced, to make it easier to build systems using such technology. In fact, IPMX takes ST 2110/NMOS as a starting point, so we can expect a high level of interoperability between devices that work in an IPMX environment and devices that work in a ST 2110/NMOS environment. However, even there, the specifications for IPMX are still being written right now, and it will then need to go through a development and initial implementation cycle.

“By contrast, if someone had an application where they do not need a lot of synchronized video and audio in multiple feeds, then they would probably use NDI, which is more of a plug-and-play implementation. But you can only go so far before you run out of steam with something like that, when you start realizing you do need to synchronize different feeds together, and you can’t do it all at the receiving location. In that case, you will need to interact more with ST 2110 and NMOS environments. So, it takes time to figure out how to do that for all these different [applications].”

Another ingredient mixed into this soup is the fact that major cloud providers are coming up with their own approaches for moving uncompressed video, audio, and data around in the cloud for various kinds of specialized production tasks. Gilmer points to Amazon’s AWS CDI (Cloud Digital Interface) networking technology, designed to help customers build live video workflows with low latency, as a prime example. In the case of those kinds of products, he suggests, high efficiency for cloud-based work is the goal of the providers first and foremost, but not necessarily interoperability, and so, how to neatly slot all of those approaches on top of an ST 2110/NMOS foundation is not necessarily a simple proposition.

“For those large cloud vendors—video is a huge amount of their traffic, and a large amount of their business,” he explains. “So, if you pick a cloud vendor like AWS or Azure or Google, they are going to be focused on the most efficient way to develop a technology that services their specific customers in a cloud environment. They are incentivized to develop efficient protocols and applications that allow them to have features that let them do things faster and better than their competitors. Therefore, the need for interoperability—whether cloud-to-cloud or even cloud-to-ground—is a requirement that might be lower on their list compared to having very efficient [proprietary] APIs to serve content consumers.

“That’s why we are now having ongoing discussions with Amazon and others about participating more in SMPTE, AMWA, and VSF activities. This is an area where I expect to see ongoing technical discussions to find areas where standards’ bodies can help make things work better, but in an open way. However, they may not necessarily want to implement the sort of open interfaces that are particularly optimized for professional media applications—after all, we are not their primary customers. Still, one way or another, we will have to figure out how to make what we do on the ground compatible with what is going on in the cloud. That is some interesting work we will see in [upcoming months].”

Michael Goldman

Related Posts