I’ve been a broadcast consultant for over 13 years, and in that time, I’ve been asked countless questions by broadcast facility customers. Rarely has there been a question that has generated so much debate.
Should I build an end to end IP-based broadcast facility?
Well, the answer is, it depends.
Firstly, what are you hoping to achieve by doing this?
- File-based workflows?
You can achieve this without going IP end to end.
- Easier for multi platform?
Not really, provided you already have a file based workflow.
- Future proof platform?
Against what? 4k? 8k? Think of the bandwidths.
- Surely, generic IP equipment must be cheaper than dedicated broadcast kit?
Not when it comes to network core switches.
- Easier to adopt cloud, SaaS and virtualisation?
All of these are good, but none of them are dependent upon a totally IP-based infrastructure.
- Simplified switching, with redundancy?
To a degree, yes, but what are you switching, and how much?
The Big Broadcast Survey 2015 identified that 46% of respondents identified IP networking and content delivery as important to the future of their business.
Recently, there have been several articles in the trade press about IP-based production, studio demonstrations and a few small-scale examples of broadcasters using IP in their core facilities.
I am in no doubt that it is currently possible, and before long will be preferable, to build completely IP-based broadcast facilities. At Media Asset Capital, we’ve even designed a couple of IP-based broadcast facilities in response to customers’ requests, but in each case, the current cost outweighed the operational advantages.
There are several issues to be tackled when dealing with IP in a broadcast environment. The first of these is that broadcast digital video is high bandwidth. 1.5Gb/s SDI (SMPTE 292M) and 3Gb/s SDI (SMPTE 424M) for uncompressed HD signals, 6Gb/s SDI (SMPTE 2081) and 12GB/s SDI (SMPTE 2082) for UHD-1 (2160×3840, 10-bit, 4:2:2 at 60 fps).
Currently, network connections are typically 100Mb/s, 1Gb/s, 10Gb/s, 40Gb/s or 100Gb/s. 25Gb/s is due to be delivered in 2016. So currently, you need 10Gb connections for HD and 40 or 100Gb connections to support UHD.
Running multiple 12Gb/s high frame rate signals through a network router will require massive backplane bandwidth. For example, a modest sized HD video router of, say, 40 x 40, provides a potential switching bandwidth of almost 120Gb/s. At 4k, this increases to almost 500Gb/s.
A Cisco Nexus 7700 switch that handles this bandwidth (and 80 ports) is going to cost upwards of $200,000 plus a significant amount of commissioning. A 12G video router of this size can be bought for only $5,000 and just plugs in (OK, so you might not want to run your plant on a $5k router, but you get the picture).
You might decide that you don’t need the ability to route every source signal to every destination, or to have all destinations available at once, and this would reduce the internal bandwidth requirements. But decide later that you do need these capabilities and it’s an expensive upgrade or replacement process.
And remember, that in order to get video signals through a network router, they need converting to IP streams. That can cost several thousand dollars per connection (80 needed in our example above).
Of course you can mitigate some of these issues by compressing the video signal. JPEG2000 and TICO offer visually lossless ‘light’ 4:1 compression, but then you need encoders and decoders. These will set you back around $1-3k each time you want to go from video to IP, or vice-versa.
And even with 4:1 compression, you still have a lot of bits flying round the network.
That’s not to say that there are not advantages in migrating to an all-IP infrastructure. It is easier to reconfigure the infrastructure as things change or the real use cases emerge over time, without the need to put in additional infrastructure while other bits sit idle, and if the equipment we are using in capture, editing, processing and playback provides standardised video IP connections then we wouldn’t need so many interfaces or encoder/decoder pairs.
What’s needed is the creation of a set of standards that are as ubiquitous in broadcast as SDI, with standardised connections on each bit of kit.
SMPTE has published several standards for broadcast IP, and both ASPEN, originally developed by Evertz, and the more recently announced Alliance for IP Media Solutions (AIMS) are great steps forward in this regard, but are in their early stages, and these two initiatives are not currently compatible.
SDI is so embedded within broadcast facilities and workflows that it cannot be swept away overnight by IP (despite the protestations of equipment vendor Cinegy that “SDI is dead”).
For the majority, the transition to IP will be gradual, starting with IP islands floating in a sea of SDI. Over time, we can expect this position to reverse to the point where the sea is IP and the islands are SDI.
SDI is far from dead. To (correctly) quote Mark Twain, “The report of my death was an exaggeration”.