In a recent post by David Ross, he extols the virtues of SDI vs IP in some scenarios.
His observations are spot on.
IP is not the defacto solution for all projects. As engineers, we still must elicit requirements (remembering that commercial requirements are ultimately technical requirements) and design the most appropriate solution based on those requirements.
As a media industry consultant, I have many discussions with broadcasters, suppliers and system integrators.
Having designed and built a big IP facility (Bloomberg London), consulted on many IP projects, contributed extensively to vendor roadmaps and written on the topic, I’m fortunate to have had a good level of involvement in the transition to IP.
Having lost my fair share of sleep from being an early IP adopter, I’m keenly aware of the realities of building an IP based facility, the costs, risks, viable architectural patterns and appropriateness of the technology to different scenarios.
David Ross’s comments resonate with my own experiences.
There’s a lot of good work going on but also a lot of poor information out there right now. Much expertise is required to architect a big IP platform and the risk of failure is high if the architecture and project approach aren’t right. Standards familiarisation is time well spent but rote learning does not deliver complex programmes of work. Big teams of well organised, obsessive compulsive engineers do. Architecting complex systems and building multidisciplinary, multi-organisational, geo-distributed teams to deliver complex systems takes experience and is fraught with subtle risks. And this is what is required right now to build big with IP.
We all have a responsibility to continually seek out assumptions about our level of knowledge and challenge them. It’s the unknown unknowns which bite hard.
A recent discussion with an Executive at an SI highlights the problem:
“We’ve recently built a small system on IP, why would anyone do anything else?”
“Cost? How much of an uplift do you think IP requires relative to SDI?”
“I’d say about 20%”
That’s a big statement. Treat it with caution. IP carries a premium and +20% is optimistic. Whatever the uplift, there are use cases where the additional cost and complexity of IP is justifiable and others where it is not.
Then there’s the basic question: What does that 20% actually include? For instance:
- Cost of fibre patching infrastructure.
- Routing control licencing (which is currently very expensive).
- Uplift in cost of professional services.
- Cost of QSFP/SFP+ modules (switch and endpoints). Those PSM4 and 25Gb modules aren’t cheap.
- Cost of additional gateway conversion.
Scale is important. Facility and multi-site deployments are very much more complex to realise. The network design approach for a big facility is nothing like a small one. Not to mention the fact that the audio side of ST 2110 is still settling both from the conformance level support and control perspectives. That dream of endpoints being able to receive multiple audio groups from different senders and shuffle them isn’t there yet.
Beware sweeping statements and the assumptions they’re built on. There are some big success stories, delivering real value, which have solved a lot of technical problems. However, we’re not quite at the stage where you can pick up 2110 capable equipment from lots of different vendors, put it on a network and expect it to work in the way that you need it to.
So what of SDI?
I challenge anyone not to be impressed by the Ross Ultrix and immediately start seeing use cases for it. In a shallow, 5RU box, you receive audio routing, 12G video routing, multiviewers and signal processing capabilities. All in an affordable, low power device, with a tiny footprint. It’s a great piece of engineering which will do well (just sort out the control profile restrictions Ross, two profiles aren’t sufficient).
For some smaller, more time and cost sensitive builds, there are clear benefits to a 12G capable SDI core. SDI is also a winner for monitoring chains. You might want to consider it for some ad-hoc long runs too:
“I need to send one signal from A to B. ONE SIGNAL. It used to be that I needed one fibre for this, now I need four”
“And the optics”
“Jesus and the optics. I’m sorry Grant, this is s*#te”
There’s nothing like a grouchy, gifted and vocal studio design engineer to keep you grounded. Live production crews don’t care about IP unless it brings tangible benefits. As an industry, we’re about media production and distribution. IP is just another tool in our armoury, albeit a powerful one.
I’m a great fan of IP and I’m thoroughly enjoying this transition. There are good reasons for it in many scenarios but it is not the defacto solution for all signal exchange requirements. For many use cases, there remain stronger arguments for SDI and 12G SDI reinforces that.
And remember, when you do build a new IP platform, there will be many scenarios where SDI remains the only right decision.
Link to David Ross’s post: