DigitalWell™, is directed at streamlining and simplifying the acquisition, aggregation, cataloging, storage, and distribution processes associated with managing large collections of digital content.  DigitalWell™ is essentially a multimedia data warehouse supporting a diverse set of digital sources and formats.  The system is designed to support clustered pluggable components allowing the architecture to scale, recover and evolve with new technologies over time.

DigitalWell™ is comprised of three main components:

1. Snapshot of the ResearchChannel tree inside DigitalWell™ (enlarge image)

   Administration Tools: These web based tools allow users to manage their digital collections. These tools feature the ability to add content in a hierarchical structure complete with extensive ACL structures that govern the use of the content. The tools allow the user to add their own custom meta-data fields on top of the Dublin Core base schema.
   
   They also allow large file, 4+ gigabyte, upload through the web interface as well as other ingest methods. A batch upload capability allows the user to upload directory structures with an attached XML document describing the collection. The batch process with then create the folders and meta-data objects automatically.
   
   

2. Example of website using web service API to display content (enlarge image)

   Web Service API: The web service API allows users to programmatically retrieve meta-data and file pointers from their collection. Using SOAP calls users can build websites that can display their content. Because we use permanent identifiers that are built into this services the correct file pointers will always be retrieved.
   
3. System Administration Tools: The system administration tools allow the administrators of DigitalWell™ to perform all the maintenance and update tasks within any given instance.

Storage and Streaming Architecture

While the components listed above sound simple, there are a lot of things going on behind the scenes to make DigitalWell™ possible. Our architecture was developed with the plug and play philosophy. We can scale linearly to keep up with demand. Each of the three main services is powered by a load balancer that can take web servers in and out of rotation based on service conditions. This is the same case with our on-demand streaming environment. New streaming media formats can easily be added and scaled as demand dictates.

Here is a basic diagram of a user request to any of the three services:

The storage that powers DigitalWell™ works in the same fashion. We incorporated the idea of storage pools into DigitalWell™. This means that individual collections can be mapped to different storage sources. Our primary storage has snapshots taken of it nightly for disaster recovery purposes.

Streaming Services and Status

DigitalWell™ is currently in production supporting live and on-demand programming for ResearchChannel, UWTV and KEXP.  Supported streaming formats include locally UW developed  Internet iHDTV, broadcast quality MPEG2 and MPEG4 (IBM VideoCharger), Windows Media, Quicktime, and Real Networks.  Available bit rates span everything from low end modem speeds to uncompressed audio at 1.4Mbs and uncompressed HDTV at 1.5Gbps.