Caton Technology – Cover the globe with internet connectivity
Michael Yang, Caton Technology
Media is now a global business. Audiences anywhere are clamoring for content from everywhere.
The K-Pop phenomenon means that a concert taking place in Seoul can attract a huge audience in Seattle and Sienna. In recent weeks sports fans globally have been gripped by world championships: cycling in Scotland; netball in South Africa and football in Australia and New Zealand.
Media connectivity is more than just television coverage of sports or concert relays to theaters.
Like every other aspect of media technology, once we were limited by expensive, bespoke hardware: now the tools for production and delivery are freely available to all. It is the connectivity that has struggled to keep up. How do we get signals from wherever the event is – football stadium or operating theatre – to wherever the audience is?
Traditionally, this meant satellite links. However, the bandwidth capacity is limited and very highly priced. Dedicated fiber lines may be available to premium locations, but the essence of new agility is the need to connect any location to any point of delivery.
There is a fabric which does now connect everywhere to everywhere: the public internet. The problem is that the internet is a wild, unmanaged environment that is unreliable and unpredictable for the performance and quality requirements of the media industry.
We are all aware of the problem of finding our internet connections going down. Network failures are very common. Cisco has provided a live tracking website, thousandeyes.com, which shows where all the major outages are, in real time. As many as 300 major network failures are seen every week, with close to half in the United States alone.
Each link in the internet is provided by a telco, and it is in their interest to move the signal on to the next telco as quickly as possible – the “hot potato” principle. But where there are failures, or data traffic bottlenecks, then the ISPs are forced to reroute streams over longer, and therefore slower, paths.
At Caton, we have developed tools to trace the paths between all the peering nodes touched by a data stream and found that latencies can vary by a factor of as much as 100 times. This is completely out of the control of the typical user. For a broadcast engineer, expecting very high reliability, no jitter and stable latencies, this is difficult to accept.
All these issues pose major challenges for delivering high quality, high bitrate media streams across the public internet. On the other hand, it is widely and readily available: virtually every location now has broadband internet access. And it is potentially very low cost.
But as the saying goes: there is stable and reliable; there is fast to implement; and there is cheap. You can have one, maybe you can have two. But having all three is tough.
Modern protocols like SRT and Zixi are great at delivering the content, but they are at the mercy of the network, simply having to accept the unreliability, latency, and capacity bottlenecks.
What is needed is a network architectural level solution that provides efficient management of the network as well as the individual streams. One that probes all the way along the routing, constantly checking network availability to ensure that the media streams are always on the best performing links with the right capacity.
Ideally, the algorithms in the network probes should be fine-tuning the route for reliability, latency, stability, and cost, to ensure that the end-to-end circuit not only delivers the program but does so for the lowest possible cost.
To achieve this requires AI powered machine learning and big data modeling, to build and maintain a network map in real time, and to make proactive rerouting decisions that are efficient, effective and instant. That is what we have done with the Caton Media XStream.
It uses our own global Caton Cloud architecture to track signals from the origin, over the public internet through our mesh network of points of presence (PoP) (we currently have more than 100), to the final destination or destinations.
Through its AI dynamic network management, it is capable of making switching decisions in less than 20ms to guide traffic through the most optimal paths. Those decisions are made for security, availability, latency, and cost.
The result is that we can offer six nines of reliability: 99.9999% up time for high bitrate transmissions for live Ultra HD and beyond. It is low cost because what we operate is an overlay core network based on the public internet infrastructure. And it is very agile, being capable of deployment virtually anywhere in a very short space of time: typically hours from enquiry to connectivity.
We know that the internet is unreliable. We cannot change that. What we can do is take a view across hundreds of mediocre, unreliable links to create one extremely reliable superhighway, through the use of AI dynamical switching to ensure the signal always gets to its destination, error free, jitter free, with minimal delay and minimal cost.
