Is the industry successfully delivering what we need with SMPTE’s ST 2110 suite of standards? What are the benefits of IP and how can we tackle the difficulties?
In this panel from Broadcast Solutions’ Innovation Day, we hear from 5 vendors understanding their perspectives and plans for the future. Claus Pfeifer from Sony say they have now 60 sites up and running in IP. Lawo’s Phil Myers follows up saying “People know they have to go IP, it’s a matter of when they go IP.”
Whilst this is a positive start, the panel moves on to talking briefly about difficulties implementing SMPTE ST 2110. Jan Eveleens from Riedel points out many of the issues will go as we are waiting for technology to catch up regarding CPUs and bandwidth. We no longer have the same processing issues we used to for audio. Similarly with video, technology will improve and remove many of the challenges. Phil Myers feels that cloud implementation issues are not a large problem at the moment as he sees a move to bring equipment into private clouds rather than public. This way they are doing ‘remote production for buildings’.
After each vendor outlined their future plans for IP, Zoltan highlighted that IP allows NDI to co-exist with ST 2110. Many may want to use 2110 for high end sports, for others NDI fits well. Then panel felt that a concerning area of IP is the worry of how to fix problems. The knowledge level is different from country to country. So vendors not only need to work on education about IP, both for NDI and 2110, but they need to do this in a focussed way for the different markets.
As the panel comes towards the end, Claus feels that the industry started to talk too early about pure technology. “Did not discuss enough about the business benefits.” he explains such as remote production and more efficient use of equipment – avoiding ‘sleeping Capex’. Installing IP makes a lot of sense for large-scale systems. Recently broadcasters have been working at a scale requiring much more than 1024 squared routers roughly where SDI routers top out. But also, these large systems tend to have a life of over 10 years. Faced with SDI development, particularly in routers, is slowing down or stopping, for these long-lived systems it makes much more sense to use IP.
Moving video production to IP has been ongoing for over 5 years using both SMPTE ST 2022-6 and now ST-2110 but we’re still in the ‘Early Adopter’ phase, explains the Willem Vermost speaking at SMPTE 2019. Willem is the EBU topic lead for the transition to IP-based studios and he is tracking the upcoming projects with public broadcasters.
Willem talks about what’s motivating these Early Adopters. In general, he explains, they have a building move project and they are faced, as CBC (Canadian Broadcasting Corporation) was, with being the last to install an extensive SDI infrastructure – and be stuck with that for 7, 10 or more years to come – or the to be one of the first to use IP. Increasingly, they can’t justify the SDI workflow and IP, for all its risks and uncertainties, is the way forward.
CBC/Radio Canada needs to be ‘on air’ in 2020 so they put in a place a risk mitigation plan to test all the equipment before putting it in. Willem outlines what this test plan looks like and what it covers: AES67, ST 2110-40,-7, -30-, -20, EBU r148 security etc. Testing was also brought up by the BBC’s Mark Patrick when he discussed his work in bring in the BBC’s Cardiff Square building on-air. They found that automated testing was key in project delivery so that testing was quick and consistent to ensure that software/firmware patches were correctly accepted into the project.
Willem talks us through the EBU’s famous Technology Pyramid which shows to what extent each of the technologies on which media-over-IP requires has been defined and adopted by the industry. It shows that while the media aspect has been successfully deployed, there is a lot to do in, for example, security.
Difficulties arose due to different interpretations of standards. To aid in diagnosis of such issues, the LIST project has created a 2110 analysis tool and other related tools. This is created within the EBU and Willem highlights some key parts of what it does. He then shows how that connects in with the automated test programs and explains the underlying structure of how the software is built.
The talk finishes with mention of the JT-NM test plan, a summary and questions lead by Arista’s Gerard Phillips.
RIST solves a problem by transforming unmanaged networks into reliable paths for video contribution in an interoperable way. RIST not only improves reliability through re-requesting missing packets, but also comes with a range of features and tools, not least of which is tunnelling. Cobalt Digital’s EVP of engineering, Ciro Noronha explains how the protocol works and what’s next on the roadmap.
Ciro starts with a look at the RIST Simple Profile covering the ARQ negative acknowledgement (NACK) mechanism, link bonding and seamless switching. He then moves on to examine the missing features such as content encryption, authentication, simpler firewall configurations, in-band control, high bitrates, NULL packet extraction. These features define RIST’s Main Profile.
Tunnelling and Multiplexing is a technique to combine Simple Profile flows into a bi-directional tunnel, providing simpler network and encryption configuration. Using a GRE (RFC 8086) tunnel, RIST provides a full, protocol agnostic tunnel and a UDP-only reduced overheard mode which only requires 0.6% data overhead to implement. Ciro explains a number of setups, including one where the connection is initiated by the receiver – something that the Simple Profile doesn’t allow.
Authentication and Encryption are covered next. DTLS us the UDP implementation of TLS which is the security mechanism used on secure websites. This provides security to the tunnel so everything which travels through is covered. Ciro explains the pre-shared key (PSK) mechanism in the Main Profile.
The talk finishes by covering NULL Packet removal, also known as ‘bandwidth optimisation’, header extension which extends RTP’s sequence number to allow for more in-flight packets and questions from the audience.
Open source software can be found powering streaming solutions everywhere. Veterans of the industry on this panel at Streaming Media West, give us their views on how to successfully use open source in on-air projects whilst minimising risk.
The Streaming Video Alliance’s Jason Thibeault starts by finding out how much the panelists and their companies use open source in their work and expands upon that to ask how much the support model matters. After all, some projects have paid support but based on free software whereas others have free community-provided support. The feeling is that it really depends on the community; is it large and is it active? Not least of the considerations is that, in a corporate setting, if the community is quick to accuse, is it right to ask your staff to go through layers of ‘your a newbie’ and other types of pushback each time they need to get an answer?
Another key question is whether we give should back to the open source community and, if so, how. The panels discusses the difficulties in contributing code but also covers the importance of other ways of contributing – particularly when the maintainer is one individual. Contribution of money is an obvious, but often forgotten way to help but writing documentation is also really helpful as is contributing on the support forums. This all makes for a vibrant community and increases the chances that other companies will adopt the project into their workflows…which then makes the community all the stronger.
With turn-key proprietary solutions ready to to be deployed, Jason asks whether open source actually saves money on the occasions that you can, indeed, find a proprietary solution that fits your requirements.
Lastly, the panel talks about the difficulty in balancing adherence to the standards compared with the speed at which open source communities can move. They can easily deliver the full extent of the standard to date and then move on to fixing the remaining problems so far not addressed by the developing standard. Whilst this is good, they risk implementing in ways which may cause issues in the future when the standard finally catches up.
The panel session finishes with questions from the audience.