The need to transcode video and audio material from one format to another has been with us since the early days of video recording. The process of carrying out transcodes, though time-consuming, has always been rather straightforward from the operator's point of view; content would be moved between systems by playing it out of one device in one format and rerecording it into another device in a second format.
Today, the transcoding landscape has changed. In one sense, it is now a more simple process in that most content is digital and file-based, so the lengthy playout step is not necessary. Often, transcodes can be done in faster than real time. Unfortunately, there is a new complicating factor. With the proliferation of new distribution channels — the Web, IPTV and mobile phones, for example — there has been a corresponding explosion of new formats. The end result is that while transcoding can be done more easily, the sheer volume of transcoding strains existing infrastructures.
With so many formats, wrappers and third-party applications, intensive transcoding is something that must happen in practically all broadcast operations. The problem is that transcoding, like most media processing tasks, requires a tremendous amount of computing power and often requires manual intervention. This places an undue drag on the efficiency expected from a file-based workflow. The more distribution platforms you wish to support, the more workflow duplication you have. Efficient transcoding requires an advanced computer architecture.
ProXchange is a transcoding system for the Omneon MediaGrid active storage system that is designed to convert media between multiple formats. The application takes advantage of the storage system's advanced, object-based architecture to enable high-speed, high-performance processing. The key to the speed and performance is that the sophisticated, grid-based infrastructure of the MediaGrid enables the burden of individual processing jobs to be divided across multiple available servers.
Another advantage of the application is that it performs transcoding on media while it remains in place in central storage, eliminating the need for standalone systems and time-consuming transfers between systems. Additionally, the transcoding system minimizes the need for human intervention because users can configure its rules engine to convert specified content automatically from one format to another. The increased performance, simplification and streamlining of the workflow increases overall efficiency.
The transcoding system uses the additional processing speed obtained through grid computing, which enables applications to run across a pool of individual servers efficiently using hardware resources to yield improved application performance. With the use of grid processing, ProXchange can transcode multiple content streams simultaneously at speeds faster than real time.
The basic storage building block of the system is the ContentServer, which houses and manages multiple disk drives. ContentServers are X86-based servers that run standard Intel processors and have a specific amount of system memory. These resources are used to manage the disks and file slices in storage, but there is, in fact, more power available than what is required purely for storage management. By using a software Applications Programming Interface (API), applications can take advantage of the spare processing capacity that resides in the ContentServers.
The main advantage of building a transcoding application on a grid processing framework is that it affords the ability to use multiple processors in parallel, squeezing better performance out of them. In addition, grid computing methodology means more resiliency: In a grid-based system, a transcoding task can be completed even if a computing node is lost.
The ContentServers operate by connecting right to the grid network, accessing content directly. There is no need to move files to a dedicated content-processing server. This results in reduced network traffic and saves time and bandwidth. In fact, the grid processing framework ensures that network bandwidth is never a limiting factor on transcode performance.
More workflow efficiency derives from the establishment of rules for automatic conversion of media on the grid. Once a set of rules has been applied to a specific directory in MediaGrid, the transcoding system automatically begins processing any files appearing there. The system can produce multiple output formats and control parameters like bit rates and output directories. For example, it can take MPEG-2 50Mb/s I-Frame content, automatically create a transmission copy in MPEG-2 12-Mb/s Long GOP, and at the same time create an H.264 clip for distribution to IPTV or the Web. Full monitoring and reporting informs users about the progress of their jobs.
Transcoding is often something that is controlled by a higher level application like media management or automation. With this in mind, Omneon has made available a comprehensive API that allows third-party applications to take advantage of all of the ProXchange functions. This API is also presented via XML (eXtensible Markup Language) over HTTP (HyperText Transport Protocol) such that integrated applications are not limited to integration from a particular operating system or programming language, nor do they require custom libraries. In addition, the transcoding system is fully integrated in to the Media Services Framework, making all transcode functionality available via a Web services interface.
Finally, because the system processes media within the storage environment, it is by definition well integrated, requires only a modest footprint, and is cost-effective to implement and maintain.
With the explosive growth of both file-based workflows and new media formats, the need for transcoding and the attendant high levels of processing power will continue to grow. Keeping up with increasing transcode requirements while maintaining workflow efficiency requires powerful computing performance. ProXchange takes advantage of the performance offered by grid computing to meet the challenge, providing faster content processing; simplified and streamlined workflow; enhanced resiliency; and cost-effective implementation, operation, and maintenance.
Simon Eldridge is the product manager for Omneon.