Since video on demand (VOD) was first commercially deployed nearly 10 years ago, it has been extended to create a whole range of services.
The very first deployments only offered movies on demand on a rental basis. Then premium channels were added to provide a so-called subscription on demand service. Next, free on demand and broadcast on demand offerings were added. Many operators are adding time-shifting capabilities, such as Time Warner Cable’s "Start Over" service. Ultimately, network personal video recording (nPVR) services could make digital video recorders (DVRs) obsolete if contents rights issues are resolved.
VOD has clearly become a tremendously successful service, with operators routinely offering a choice from thousands of titles of standard definition (SD) programming, refreshed with significant installments of new material every week. In addition, the number of titles available in high definition (HD) has grown rapidly in recent years.
Each on-demand title comprises title metadata and a set of assets. Each asset contains content files and metadata describing them. Examples of assets are trailers, movies, poster-art, and so on. All of the assets have to be provisioned into the video server before a title can be offered to the subscriber. Title metadata and poster-art assets are used to generate the on-demand guide listing that enables the subscriber to make a selection.
Content providers create the on-demand assets and distribute them to each VOD system. Each system may have a different storage capacity, so some assets may be filtered. There is often a need to modify the metadata, and the assets must be organized into catalogs according to the operator’s requirements. This process is called "asset management."
All the above translates to tens of thousands of files, with a total market value measured in millions of dollars, being continually transmitted to and reliably managed at hundreds of server locations, creating a tremendous challenge to keep everything in a perfectly operational state.
What is more, as consumers become interested in watching their content on PCs, game consoles, portable media players and mobile phones, operators have an opportunity to differentiate their services by supporting these multiple screens. Although VOD uses an MPEG-2 transport stream in a quadrature amplitude modulation (QAM) channel to deliver video to a cable set-top box, nearly everything else about the service is independent of the delivery mechanism. Why not enhance existing systems to support "multi-platform" service delivery? To do this, we need to deliver content to multiple platforms, in multiple, device-specific formats equating to many more assets.
Make that hundreds of thousands of files to potentially thousands of server locations!
There is also a trend toward an ad-supported model for on-demand delivery, as promoted by "over-the-top" providers such as Hulu, Joost and Babelgum. Although operators may find some subscribers are willing to pay for ad-free content after the initial thrill of these competitive offerings wears off, many other subscribers will be receptive to advertiser-supported programming. Therefore, the ability to support dynamic ad placement, according to the SCTE 130 standard, will be a key consideration for operators.
Some of these new screens are also much better suited to interactivity than the TV set, being equipped with keypads and pointing devices. So any solution should support management of interactive assets, such as scripts, widgets and applets.
I’ll use the term "content management" to describe a superset of asset management that includes a broader transformation of the content, including metadata and content files required for multi-platform service delivery and interactive applications, and that supports seamless integration with dynamic ad placement systems.
Next month, I will talk more about the requirements for multi-platform content management solutions, capable of scaling to 100,000 titles or more. But first, let’s examine the history that brought us to the current situation in a bit more detail.
In the earliest days of on-demand, content files were shipped around on tape, and metadata was manually entered at each system. In order to scale on-demand services, the concept of a package was developed. This is essentially a collection of assets and metadata files that completely describes the on-demand title. Each package is transformed into a single file using a UNIX utility called tape file archiver (TAR). These TAR files are transmitted over satellite, using multicast to greatly reduce bandwidth requirements. Thousands of hours of content are distributed to systems via satellite from multiple content providers. Although an hour of SD content requires about 1.7 GB to be transferred, the actual transmission rate from each content provider is relatively low, on the order of 10 Mbps.
Unfortunately, bundling the content files with the metadata means that the metadata is also multicast, forcing a one-size-fits-all approach. This forces the operator to make metadata changes at each system, after content distribution, rather than centrally. As a result, asset management has become complicated and time consuming because each VOD system is managed separately. Worse still, there is a lack of transparency to corporate marketing and central operations groups. If delivery of a title to a site is unsuccessful, a manual "re-pitch" is required, which is both costly and time-consuming. Finally, reporting is unreliable and outdated.
A real-world example: A content provider promoting a weekend of movies featuring actor Nicolas Cage accidentally sent out one of the titles with an error in the metadata. The actor’s first name was entered incorrectly as "Nicholas." Since the metadata and content are atomically bound together, the only way to replace a single field in the metadata was to re-distribute the entire 3 GB package to all sites.
The title metadata includes fields on pricing, categorization, and availability window. A content provider distributes packages with these values set according to its business rules. The operator may need to override some of them, such as different pricing on a system-wide or system-by-system basis. Some assets may need to be filtered because they are not required at the particular system or because there is insufficient storage capacity for them. Because of the need to change the metadata for each operator, and even down to the specific site, asset management systems were introduced. These systems were deployed on a site-by-site basis initially, but have evolved to support multiple sites in a regional environment.
After the package is received at the system, it is provisioned in the back office. Part of this process is to propagate the content files to each video server that will be used to deliver streams for this title. Finally, the title is ready to be offered to the customer via the program guide.
The rest of the story
Next month, I’ll examine the requirements for an end-to-end, multi-platform content management solution to allow operators to:
• Manage content from multiple content providers, arriving with different metadata formats
• Manage incoming content via multiple ingest methods
• Achieve visibility and control of content from ingest point to each video server location
• Centrally monitor asset status across many distributed sites
• Easily and centrally change content metadata
• Centrally manage content
• Support locally created and ingested content
• Support dynamic ad placement
• Support interactive applications
Michael Adams is vice president, systems architecture, for Tandberg Television. Reach him at firstname.lastname@example.org.