Sponsored guest post by Sven Freudenfeld, Communication Business Unit, Kontron.
Video is not just video, an operator could be offering up a quick clip of streaming video from a specific application like YouTube, or a conversational video from its own or a third party service, a good quality stream of a TV programme, or a download of a film. Each of these streams have different quality requirements, must be delivered over differing access technologies to devices with different screen sizes, resolutions and flash memory sizes. The in-app short YouTube clip needs to be delivered there and then (users hate buffering), but the quality of the video may not matter so much. The video conversation or conference requires low latency and excellent audio quality, the stream requires good bandwidth and the download may even be better being offloaded to a different access technology, if the user is sitting on a sketchy HSPA connection in a contended cell.
A video optimisation platform does all this. It transrates video streams from source quality to output video at the appropriate quality for the connection and device. It transcodes video streams per device, and it may also apply operator policies that are based on quality parameters and user privileges. This is what platforms currently do. The issue for the providers and operators of these video optimisastion and monetisation platforms is that video – all of video – is a moving target. Today’s platform is only good for so long: devices, applications, even networks are changing at high speeds. The volume of video demand, and the number of sources are also rapidly increasing. Current forecasts are that nearly 70% of all data traffic will be video by 2017. Some operators have reported that LTE traffic volume has been as high as 90% video.
So the management of video traffic will be key to the operational efficiency, network performance, and revenue protection and generation of mobile networks. That means that the performance, functionality and scaleability of video optimisation platforms are crucial.
This is why we have seen the introduction of cloud-based video platforms – with their offer of load-balanced and the dynamic availability of resources turning up and down to meet demand. Cloud looks to have a key advantage in the management of video traffic. We’re going to see more video optimisation and management, more analytics and data processing. We’re going to see more of that happen in environments that can take advantage of cloud management and orchestration such as shared resources and dynamic scaling.
Just as video is not just video, not all cloud platforms are created equal.
Yet just as video is not just video, not all cloud platforms are created equal. With talk of software or virtualised images running on industry standard IT platforms it’s tempting to think that one cloud platform is much like another. That’s not the case.
So what should you look for in your video optimisation platform?
A platform must offer you the availability, scaleability and functionality you require at a cost that is much more efficient than scaling current platforms to peak demand. Using a platform designed for the cloud that offers the highest channel density per CPU thread will offer greater processing capacity by performing transcoding functions on the GPU. That also leaves headroom for performing revenue generating functions such as analytics on the CPU. Such a system can transform the economics of video optimisation, reducing the power per video steam by as much as 75%, whilst increasing performance and functionality in a radical fashion.
For example, five SymKloud platforms from Kontron, based on fourth generation Intel Core i7 processors, can transcode up to 1,080 MPEG2 streams to H.264 – something it would take 17 blade centres to achieve. Add to this a tremendous power savings potential up to 50K Watts.
Mobile operators faced with the challenge of efficiently delivering video over their networks can take advantage of these new archtectures, based on the latest processor platforms, to not only handle current volumes of traffic more efficiently, but to deal with future volumes in a manner that frees up other network resources, and offers the opportunity to derive revenue generating services from deep analytics – all performed on the same platform.
For more on Kontron’s SymKloud Media platform, based on fourth generation Intel Core i7 processors go to SYMKLOUD.com for whitepapers and other resources.
Kontron will feature live cloud demos during Mobile World Congress, at Stand 5H19, Hall 5.