IP Production Forum: Virtualized, SaaS-Based Operations Come Into Focus Agility, reliability, flexibility are the goals of computer/cloud-based workflows By Ken Kerschbaumer, Editorial Director Friday, May 11, 2018 - 1:07 pm
Print This Story
Story Highlights
Virtualized production, the ability to have a computer with multiple cores handling multiple production processes coupled with cloud services, was the focus of a panel discussion at the 2018 SVG IP Production Forum May 9 in New York City.
Mark Hilton, VP, live production products, Grass Valley, said that the goals of virtualization and SaaS (software as a service) are similar to those of the move to IP: adding agility and flexibility to a production. Depending on the application, he said, the need is greater or less. And, while a switcher control panel will still be there, a lot of the processing behind it will be virtualized.
From left: GHO Group's Gary Olson, Google Cloud's Saurabh Gupta, Grass Valley's Mark Hilton, Aperi's Joop Janssen, NHL's Grant Nodine, NewTek's Chris Burgos, and Aspera's Jonathan Solomon
Saurabh Gupta, customer engineer, Google Cloud, noted that one of the benefits of a virtualized environment is that reliability is improved because software-based cloud systems are designed to fail but remain running.
It's how you get to five or six nines of reliability, he said. The same thing applies to networking and storage, and it has changed how the media supply chain is handled.
Grant Nodine, SVP, technology, NHL, provided an example of a possible workflow that exemplifies the virtualized future: cameras with 8K resolution are installed in arenas, and cropped regions of the 8K signal are available to the production team. Like many possible workflows, his envisioned workflow is yet another example of one technology, 8K, waiting for the birth of another (8K codecs, processing, and workflows) to enable it. And that is where virtualization could play a key role in turning vision into reality.
Generally speaking, he explained, down the line, all signal handling will be in a virtualized piece of hardware and not in a stack of equipment frames with card-based solutions.
Hilton expanded on Nodine's workflow, adding that virtualized image processing will allow uncompressed live signals to be used for the core production while compressed versions of the images are made available to an editing team, which would then create an EDL that would access uncompressed images in the cloud for final full-resolution rendering.
Latency is important in live sports, he added, and the quality needed for making edit decisions is not the same as you need for final distribution.
Saurabh Gupta, customer engineer, Google Cloud, pointed out that bandwidth is a solvable problem, especially once fiber becomes the backbone.
You can capture, store, and have real-time mixing and matching of however many streams you want, he said. The bottlenecks of today will dissipate. You can have virtualized infrastructures in all of the production trucks, and they can always have the most current version of software: it will update just like the way you update software on your phone.
Aperi CEO Joop Janssen, noting that the technology is here today, said the key issue is to minimize latency so that a production-team member - whether onsite, at the broadcast center, or at a data center - has the low-latency performance and ability to spin up all the required functions.
We are at a fantastic point in time where we have virtualized low-latency production functions and have created a server with low-latency computes that are directly attached to an IP network, he said. So you can distribute those computes over your network wherever and whenever you need them. And the hardware is basically at COTS price. So that is not where the big investment is. Instead, the investment and added value are in the software and how you deploy them. If you create a SaaS model with floating licenses, you can completely optimize your production and cost on a game-by-game basis.
Control of Computing Power Helps Control Costs
Currently, there are costs associated with the bandwidth and connectivity for such functions, but the benefits of increased control over the computing power across a production enterprise are a net positive.
Chris Burgos, territory manager and workflow specialist, NewTek, added that, ultimately, the move to a virtualized environment will allow sports-production companies to make the most of a small budget.
There are costs in deploying a control room or just moving gear and trucks and power consumption, he pointed out. Virtualization will make it easier because you won't have to build out a new room, and you can cover all kinds of sports because you just spin up more virtual machines and software solutions.
NHL's Grant Nodine (left) and NewTek's Chris Burgos
IP standards and software-based technology, such as NewTek's NDI, are half the battle in the move to virtualization. Burgos suggested that, for practical purposes, virtualization for a sports production today will be limited to onsite operations because latency and other issues limit the opportunities of cloud-based operations. But the short story is that it is possible today and practical to start the move towards the cloud.
Jonathan Solomon, strategic initiative engineer, FASPStream, Aspera (an IBM Company), observed that the biggest challenge in making the move to a virtualized world is people.
People don't like change, he pointed out. Some people want to walk in and see their playout server. They don't want to be told it's one of these cards in a Cisco array. That's the biggest problem with virtualization, and it leads to the biggest problem with SaaS.
Janssen added that the change challenge i
