The tipping point
May 1, 2007 12:00 PM, BY JOHN LUFF
This year, video servers may finally replace videotape.
This year, TV recording technology may finally reach its tipping point. Ever since the first computer-based video servers were introduced at NAB in 1994, industry analysts have predicted that they would replace videotape.
Just the stats
In 2006, HD content could be stored at 50Mb/s cheaper on disk than tape, according to Screen Digest. (See Table 1.) If these estimates are accurate, the differential in 2007 will be 2:1 in favor of disk storage.
It's too early to predict the death of videotape on the basis of only one data point. To truly eliminate tape disk, or any other media, new technology must be capable of replacing the workflow that tape supports. In addition, it will be impossible for videotape to die out until it is economically efficient to replace all of the media archives stored on tape. Given the large amount of storage on tape around the world, it seems unlikely that all content will be converted any time soon.
The storage pyramid
Today's modern storage environment is often a mix of many types of storage, all picked to enhance the workflow in specific ways. Tape, along with removable disks, memory cards and optical disk storage, is the media of choice for all types of field acquisition. Using nonremovable media in the field would radically change the workflow. When hard drives reach sufficient capacity in hardened packages comparable in size and cost to other removable media, they might also be a practical option. For now, they are still much more expensive than other media for field use.
Figure 1. Storage systems usually take a hierarchical form, with media that needs to be accessed quickly held close to the point of use.
The situation in fixed facilities is quite different. Certainly tape remains viable, especially considering the impact of workflow changes in existing facilities. But with the new economics, tiered storage becomes a viable alternative. (See Figure 1.) Hard disks are almost always part of this approach. The concept of hard disks does not focus so much on how a file is stored but rather whether it arrives in a timely manner.
This is quite different from videotape, which relies more on where the physical media resides. Videotape is tangible. Hard disk enables an inherently nonlinear workflow. It also allows the use of the most cost-effective media.
At the top of the pyramid, the cost per gigabyte is highest, and the access is the fastest. Transfers can be faster than real time. At the bottom of the pyramid, access is slower, cost is lower, and capacity is larger. Changes to the storage structure can be made without disrupting existing files in any part of the pyramid because transfers to new types of media can be made without reconstituting the content itself.
Prolonging the life of content
This general approach has some issues, which professional archivists fight daily. About 10 years ago, an industry guru told me that once content was digital, it would be impossible to lose it. Unfortunately, the very fact that the content has been stored for computer access creates potential issues.
Let's say you stored content in a wavelet-based editing system in 1996. Picture quality would likely have been wonderful. It would be the same today, as long as the codec that was used could run under currently available operating systems and hardware. That, however, would be problematic.
Graphics stored on 8in floppy disks were commonplace at one time but are unreadable today. The only sure way to prevent this obsolescence is to clone the content, making a copy in a new format that will extend the life of the file for the future.
This is not as easy as a simple file copy. For some formats, such as MPEG-2, decoders will be available long into the future. Other more proprietary formats may have less longevity. The only way to protect these formats is to decode them to retrieve the original essence and then restore them using new tools.
File formats, as distinguished from compression or essence formats, can be handled with more elegant tools. Unwrapping and rewrapping can be done without changing the essence. That might be appropriate if you are moving the content from a discontinued video server's proprietary format to the house format for a new server. Converting, or flipping, the file perfectly recreates the essence in a new format.
Media archives
An analogous issue exists with media archives. SMPTE is in the initial stages of creating a standard language to speak to archives, as well as a common file format that all archive manufacturers would adopt for media content.
In the not too distant future, we should be able to move content from one facility to another in a seamless way, literally between two competing archive manufacturers. Users would then be able to order content from a service provider for inclusion in a VOD system without wondering if the archive tape it is delivered on will play in the archive they own.
MXF has a place in this continuum as well. Those who have been struggling to make MXF work in the real world know it is only a language, and it can be implemented in a spectrum of choices. At the instigation of Turner Entertainment, the Advanced Media Workflow Association is working on a proposal that will put some boundaries on a specific implementation of MXF for program storage purposes. If agreed on by industry groups, it will be an important methodology, which will help make the tapeless world more transparent.
A new pyramid
This year of transition will experience continuous downward pressure on storage costs. I look forward to the day my personal video recorder has a couple of terabytes of storage and my HDV camcorder can transfer files directly to the PVR's storage without conversion. In the final layer of the tiered pyramid, it is not unreasonable for the file to be transferred directly to storage in the home, which begins a new pyramid on top.
John Luff is a broadcast technology consultant.
| Year | Hard disk cost per gigabyte ($) | Hard disk (50Mb/s) cost per hour ($) | PRO DV tape cost per hour ($) |
|---|---|---|---|
| 1995 | 820.00 | 18,450.00 | 20.00 |
| 1996 | 256.00 | 5760.00 | 20.00 |
| 1997 | 123.00 | 2767.00 | 15.00 |
| 1998 | 62.00 | 1395.00 | 15.00 |
| 1999 | 31.00 | 697.50 | 12.00 |
| 2000 | 12.00 | 270.00 | 12.00 |
| 2001 | 6.00 | 135.00 | 12.00 |
| 2002 | 3.00 | 67.50 | 12.00 |
| 2003 | 2.00 | 45.00 | 12.00 |
| 2004 | 1.50 | 33.75 | 12.00 |
| 2005 | 0.70 | 15.75 | 10.00 |
| 2006 | 0.35 | 7.88 | 10.00 |
| 2007 | 0.20 | 4.50 | 10.00 |
| 2008 | 0.10 | 2.25 | 8.00 |
| 2009 | 0.04 | 0.90 | 8.00 |
| 2010 | 0.03 | 0.68 | 8.00 |
Send questions and comments to: john.luff@penton.com
| Want to use this article? Click here for options! |  |
