NAB Special Report: No more 'IP-pimped SDI'

Aperi at NAB 2016
Joop Janssen, Aperi
May 19th 2016 at 4:56PM : By

Aperi was founded with a vision of putting data and cloud technologies at the centre of broadcast infrastructure. No wonder that Aperi board member Joop Janssen is impatient with the industries progress.

Walking around NAB this year, it was great to see the majority of vendors offering robust and interoperable IP-connected products for most, if not all, of the broadcast/media chain. I’m also happy to see more and more facility-wide IP connected projects being installed as we speak. This is good news for the industry, which has been planning the IP migration for some years. IP’s advantages are clear, especially for greenfield implementations. But what surprised me is that most vendors, and especially the big ones, are still not offering true open, native-IP products. These are the solutions that will not only benefit from volume COTS hardware and open-source software but also from the huge advantages of true virtualised media functions. More on this later, but first the good news:

We seem to have arrived at a consensus regarding IP protocol standards with the AIMS initiative and the evolution from SMPTE 2022-1, -2, 5,-6, -7 to VSF TR03/04, which would very nicely leverage the flexibility, efficiency, robustness and performance of the datacomm world. Interoperability tests between vendors, on display at the show, showed tremendous progress. And orchestration systems available are maturing to be able to handle the migration to a full IP-centric architecture very well.

Infected by SDI

While these strides are very important, I didn’t see vendors moving beyond solutions based on an IP connector on an SDI box. Too many products labeled ‘IP solutions’ still have an SDI architecture at their core. In fact, they’re not COTS at all. They require extra translation steps. They’re fixed–function, rigid (and likely obsolete) when network and video standards evolve, which they certainly will. The most critical point to remember: most of the advantages brought by moving to IP/Ethernet architectures will not be realised if we don’t first virtualise media and network processing functions.

At its core, the challenge is to figure out how to turn what has traditionally been hardware into software and apply today’s best data centre principles. By loading all the functionality needed for live production as software via a microserver inside the IP switch you can achieve virtualisation.

Virtualisation seems like a catchall phrase today. But the key to virtualisation is really microservers and microservices – like assigning a piece of resource in a data centre to perform a specific function. I like to use the example of cell phones. We no longer need a separate camera, calculator or GPS. These functions are virtualised aggregations available when we need them. The same can be done for media processing functions but only by taking a fresh approach to how.

Customers don’t have to compromise with transitional ‘IP-pimped SDI’ products any more

The need for speed

Traditional servers, power sucking and heat generating, are also slow. GPU-based infrastructure is costly and not powerful enough for live processing. FPGA technology provides the speed necessary and is the best way to achieve the low latencies required in the high-quality live production market. FPGA technology is, however, difficult to use and program. But by extracting its complexity, packaging it creatively, and applying converging technologies of the last several years, it’s possible to flexibly configure an FPGA-based computing platform, just like a CPU. With this you can create open, IP-native and fully switcher-based solutions and move traditional hardware functions to software available through apps.

This type of approach also provides the speed and agility required to respond to today’s changing media consumption patterns. Broadcasters and content providers need to be able to spin up new channels (the OTT era) in a matter of minutes, not months. This environment also requires a shift in thinking, taking time to consider what apps are needed to get you up and running.

Consider live broadcasting where only 20-30% of expensive processing equipment is used at any time, the rest of the time it sits idle – not the most efficient or cost-effective operating model. In the software environment, you only pay for what you need, paving the way for new use models and monetisation opportunities.

IP is ready for the most mission-critical live broadcast and it was great to see new companies at NAB thinking differently for the first time. With solutions like a working, native-IP and fully software-based system with a COTS foundation, customers don’t have to compromise with transitional ‘IP-pimped SDI’ products any more.