Applied Technology: Avid's media networks

People in broadcast are talking about workflow. Workflow improvements mean productivity gains. And, in these competitive times, any significant advantage in cost or capability can have an immediate impact on the bottom line. However, news production has been slow to move from linear, tape-based processes toward nonlinear technology, which provides improved workflow. While digital video servers and nonlinear editing systems have undeniably made playout and editing operations more efficient, the promise of nonlinear technology won’t be truly fulfilled until these applications are networked together, that is, until there is an optimal nonlinear workflow. When this happens, everyone in the facility will be able to see, access and use all current and archived media instantly and simultaneously.

To support nonlinear media workflow, a digital infrastructure had to be invented. It had to be capable of supporting dozens of simultaneous high-resolution media streams in real time, and it had to have several other attributes required by broadcasters to support news production. What was needed was a “media network” — a network designed to move, manage and create digital media.

Essential characteristics

For over eight years, Avid Technology produced and delivered shared-media networks to support the workflow of nonlinear content creation. With accumulated experience and substantial customer input, Avid began developing the media network five years ago, and defined it as having the following essential characteristics:

  • Efficient shared storage — Work flows more efficiently and easily if media appears everywhere but actually exists in only one place, with the possible exception of a dedicated cache for ingested material, program or spot playout.
  • Ample bandwidth — The term “media network” implies connection. Handling multiple, high-resolution media steams simultaneously and reliably requires an extraordinary amount of bandwidth.
  • Integrated media asset management — With news production, media comes in, goes out, and expires with tremendous volume and velocity. How can you find and use what you’re looking for most efficiently?
  • Flexibility — Things change. A media network should not be rigid, proprietary or impose awkward organization. Entire volumes should not have to be locked when one person needs to record. A network should accommodate new applications and technology while supporting legacy components. It should operate with systems from multiple vendors. And it should preserve your investments, growing and adapting with you.
  • Scalability — As users are added to the network, can bandwidth expand to ensure real-time access by all? If not, media access will become limited or awkward, and will no longer be in real time.
  • Reliability — It goes without saying, but there shouldn’t be an active single point of failure. Reduced service may be acceptable, but essential operations must continue in event of component failure.
  • Usable — The system should be transparent, and administration should be complete but straightforward.

Most importantly, to deliver all of the these attributes, the data architecture must support all of the above criteria. The architecture should be independent of the base networking technology and, ideally, should allow the use of Fibre Channel, Ethernet, etc.

Today’s technology advancements address the above customer-defined criteria. Now, broadcasters have a central storage resource with simultaneous access to any media. Such systems can be configured to support over 100 simultaneous DV25 streams and 200 tracks of 48 kHz audio. Fundamental to these systems is the use of a media-optimized file system (akin to an open media file system), and a server-assisted architecture. To use the media, clients reference the location of media on storage arrays through a controlling server. The actual transfer of the media, however, does not go through the server. Instead, leading industry vendors have engineered an open-media framework that provides high throughput by efficiently striping across every drive in the system using a media-protection scheme that is a proprietary, advanced version of a RAID. Unlike the commonly used RAID3, the newest versions use block-level mirroring and striping, which actually adds bandwidth and the flexibility to switch media protection on and off per virtual partition. In such instances, the client automatically selects the media chunk from the drive that can deliver it the fastest. If one drive fails, the media will still be available. High aggregate bandwidth also provides the margin for reliable operation with multiple, high-resolution streams, so that the system has no active single points of failure.

The first practical application of such open media frameworks is “virtual storage.” This means that storage is treated as an abstract, elastic and transparent “property.” It is seen as virtual partitions or workspaces that can be created and changed at any time without destroying media or interrupting work. If more storage is needed to record a feed, it can be added from the available storage pool.

Browser-based applications address the ability to find and use media. They have the ability to automatically register media metadata when recorded into the system. An advantage of using a Web browser for such a system is that anyone anywhere can find and use the media.

In conjunction with dedicated media-transfer clients, users in remote sites can easily search each other’s materials and transfer media via existing IP networks. To transfer or edit selected material, users simply drag it into the bin in their application.

To understand what a media network is, it helps to understand what it isn’t. A media network is not a storage-area network (SAN). SANs were developed for general-purpose transaction-processing environments, and optimized to consolidate diverse storage arrays and deliver small chunks of data reliably over conventional networks.

Digital video is a far different data type, and dealing with it effectively affects every aspect of a system’s software and hardware. If you look at the defining characteristics of the media network, it’s easy to judge whether these systems are your best long-term investment.

Another thing to look at is the role of digital video servers. These purpose-built computers are great at the dedicated applications for which they were designed, such as program and spot playout. Networking servers together and touting them as “environments” that handle ingest, editing and play-to-air, however, falls well short of creating a true media network for two principal reasons. First, digital video servers were never designed to handle large amounts of storage or streams. Second, the file pushing required is inefficient, creating access bottlenecks and/or forcing wasteful media-file replication across servers. In the end, video-server-based networks don’t deliver much value for their considerable cost.

Media networks are a new kind of productivity tool; one that comes in many sizes but has several common and essential aspects. Most importantly, the media network is the fundamental digital infrastructure that can provide the workflow improvements that will drive the next revolution in news production.

Jim Frantzreb is a senior product marketing manager at Avid Technology.