Be more efficient with a loosely-coupled design.
Broadcasters are finding themselves in an ever-changing environment. Being able to meet the challenges these changes present is key to remaining relevant in a world increasingly dominated by alternative content delivery methods.
A traditional facility might include a newsroom, a production facility and master control. Each area of the facility is provided with all of the technical elements needed in that area. For example, in a server-based facility, a newsroom feed recording area might have a dedicated server. The production facility would have dedicated edit servers; the same would be the case for master control. (See Figure 1.) In each of these areas, the servers perform the same function — recording and playing back video and audio. But the server might be used in different ways in each area: recording satellite feeds in the news area, acting as a source in the post area or playing back programming in master control.
Figure 1 illustrates a traditional stove-pipe design, dedicating transcoders, servers and other equipment to specific technical areas. This tightly coupled design style makes it difficult to adapt to changes in workflow. What happens when you add new functionality to such a facility? Or what happens when management requests the capability to feed mobile devices or to serve VOD content? Of course, broadcasters are already faced with these demands, and while they have not built parallel facilities for each new distribution outlet, they have tacked on pieces here and there to meet increasing demands. This approach works, but is there a new way of thinking that will allow broadcasters to respond more quickly to the changing competitive environment and make better use of the technical assets they already have?
Loose coupling is a concept that is used frequently in workflow design. It means that the function performed by a device is loosely coupled to the characteristics of the area in which that function is performed. This is a key concept in facility design because it allows you to start thinking about common functions and processes in your facility, and it allows you to use resources more efficiently. Furthermore, it enables you to design facilities in a way that makes changing the facility easier. For more information on loose coupling and other workflow terms, see the Advanced Media Workflow Glossary online at wiki.amwa.tv.
Let's look again at the concept of video servers in the television facility. Regardless of the area in which the server is deployed, the server's primary function is to record, store and play back video, audio and ancillary data and to receive and deliver video content as files. If a well-known and publicly defined network interface can be developed, then end-user devices such as newsroom feed clients can use the same servers as editors and master control automation devices.
While fully deploying workflow-based design in a television facility is a tall order, there is no reason you cannot begin to employ this way of thinking now. In fact, the concept of a centralized tape room in post and news facilities has been around for some time. In some ways, this was one of the first implementations of workflow-based facility design. It allowed the reuse of tape machines in different areas by recognizing that the function these machines performed was identical regardless of where they were being used.
As broadcasters move to file-based workflows, one of the challenges continues to be getting the video and audio into a format that the target device understands. For this reason, transcoding has become a fact of life in most television facilities. Many transcoding devices employ watched folders. Files placed in these folders are automatically picked up by the transcoder and converted from one file format to another.
You might not think of watched folders as part of workflow-based facility design, but they are. Transcoding is a function that is needed in many places in the television facility. If you followed the old model of facility design, each area — news, post and master control — might require its own transcoding facilities. But following a workflow-based model, a pool of transcoders could be made available on the network to any user needing that service, regardless of where they are located. In fact, the physical location of the transcoders is unimportant. Because the transcoding function is loosely coupled, it can be made available to new areas, as required. Also, loose coupling allows the broadcaster to add new transcoding capabilities to the facility without major re-engineering.
The same idea can be used with video servers. The video server service is used in many places in a facility. Is it possible to design a facility with centralized video server services such that the client requiring these services is loosely coupled to the device storing the video?
Figure 2 illustrates a redesign of the facility shown in Figure 1. Note now that transcode and video server services are offered to users across the network. Also, now fewer transcoders and video servers are deployed because redundancy has been eliminiated from the traditional design.
The idea that a common set of processes are required in a number of different places within a facility is not new. In fact, many broadcasters have already analyzed their facilities, looking for common tasks across departments that are performed over and over again. In fact, the European Broadcast Union has conducted a study of common processes. This work is important because it identifies processes that can be optimized into service offerings, which can then be made available throughout the facility.
The next step in workflow technology is to move beyond watched folders and pooled tape facilities to workflow-based facilities where services such as transcode and ingest are offered on the network to clients that need them. One way to develop such a system would be to have clients communicate with devices offering services through a countless number of proprietary APIs. This would work, but it would destroy the loose coupling that is at the center of workflow-based design. Each user would have to know the make, model and API software version of the transcoder they are trying to use. This is not the ideal solution.
A better way might be to develop a common language so that the services being offered can be described as such that anyone could write a description of the services, and any client could discover and use these services, regardless of the manufacturer. This is a lofty goal, but it has been achieved in other industries.
If common processes can be identified, and if the industry can reach a consensus on a way to define and use services, then the next logical step is to use workflow orchestration tools to redefine workflow processes on the fly. Imagine a workflow orchestration system that defines a workflow (ingest, edit, approve and move to air playout servers, for example) through a workflow diagram. A manager comes into your office with a new business plan that requires making video content available on the Web after it has gone through a separate approval process. In the future, it may be possible to add this new business to your workflow by editing the block diagram of your facility, adding the individual processes required (low-res ingest, Web approval, publish to Web). This is the promise of workflow-based facility design.
Of course, this article has focused on the theoretical aspects of workflow-based design. For practical reasons, I suspect that master control will always have its own dedicated resources. Similarly, for technical or political reasons, news and post may also require dedicated resources. But it is important to realize that facilities can be designed differently from the way we have done it in the past. Doing so greatly increases our ability to respond to changes in the industry.
Brad Gilmer is president of Gilmer & Associates, executive director of the Video Services Forum and executive director of the Advanced Media Workflow Association.
Send questions and comments to: firstname.lastname@example.org