Video storage

In February's article, I discussed archival storage. This month the topic is rather similar, but nonetheless quite different.

Video compression specialist and author Peter Symes says, “Recording is transmission with indeterminate latency.” At one time, that storing of video was possible only as a film process, in perhaps the first convergence of film and television technology.

Kinescope recording was a process in which a television picture on a black and white monitor was shot with a film camera. The earliest reference I could find was to a system introduced by Kodak in 1947 called the Eastman Television Recording Camera, which was developed in cooperation with DuMont Laboratories and NBC. There was even a DuMont-developed electronic studio camera with a 35mm film camera inside.

Needless to say, these methods were developed due to the lack of electronic recording apparatus and were driven by needs, including the need to deliver time-zone-delayed programming. I am sure you have seen old programs of low quality that are often out of focus and have limited dynamic range. Experiments in commercial recording began as early as 1951, and AMPEX publicly showed electronic recording using the quadruplex 2in video recorder at NAB in 1956. In 2005, NATAS awarded a Technical Emmy for Lifetime Achievement in Technology to the AMPEX development team.

Today the needs of video recording have become enormous, and the development that AMPEX first brought to the market has become commonplace in the home, in addition to professional usage. Unless content is live, broadcasters must acquire programming by either recording or delivering recordings that were made elsewhere.

Lately, the question is whether that recording should be made on linear media (videotape) or on nonlinear media, such as servers, extended memory (like Panasonic's P-2), rotating media (like Sony's Blu-ray media) or evolving holographic storage. Another question is what the disposition of that content would be in each case — whether it should be transferred into hierarchical storage or simply put on a shelf.

Linear vs. nonlinear

The first recording decision is whether to use linear tape or nonlinear media. The opinion in vogue today is that nonlinear storage is so inexpensive that videotape can be discounted for future use. I would not yet make that assumption. While videotape does not have some of the desirable characteristics of nonlinear media, it is not obvious that the death of linear recording is upon us. It is still inexpensive, durable, portable and ubiquitous.

Think about a rack of hard disks containing 50TB of storage. That might hold a little less than 5000 hours of DV content. For the sake of argument, let's say that it fits in one rack. The volume of that same rack (78in × 22in × 30in) could hold a little more than 5000 large DV cassettes tightly packed, each of which holds four and a half hours at consumer recording speeds. At something like 4 times the volumetric storage density, it seems like linear tape is not so inefficient after all.

Of course, there are other tradeoffs. Linear tape is slow when searching content. Extracting a short clip means finding the correct tape, loading it, shuttling to the right spot, extracting the needed shots and then reversing the process. It is fundamentally different and slower than nonlinear access, especially when the mounted drive contains large amounts of content.

But putting hard drives on the shelf for storage of content is not efficient either. One drive may contain fragments of hundreds of pieces of media, and it may require an entire striped array to retrieve any useful content.

DVD, Blu-ray and HD-DVD

DVD, Blu-ray and HD-DVD offer another alternative with a modest amount more capacity than older removable nonlinear media, but the problem still exists. Unless a DVD has definable files that can be searched on another platform, it might not offer an improvement. If the content is stored as files, it must be read by an application capable of turning that file back into useable content. Fortunately, loads of applications exist for doing just this, which again borders on previous information about archiving.

Each recording technology has its own limitations. Optical recording will likely never approach the speed of hard-disk recording and replay. They have, however, achieved sufficient throughput to allow HDTV recording at consequential bit rates. They also present interesting environmental challenges that require care to overcome. Both high and low temperatures represent problems for optical media, and dust can present a particular problem.

Holographic recording

One emerging technology with great potential is holographic, sometimes called volumetric, recording. By using the full volume of the recording media, instead of a small number of layers, high density recordings and extremely fast access can be achieved. (See Figures 1 and 2 on pages 484-485.)

The principle is simple: The execution in hardware is much like the rotating media of a DVD, but the capacity of 300GB per platter and the throughput of 160Mb/s is quite encouraging. InPhase Technologies and Hitachi are releasing an initial Blu-ray product for professional archiving late this year. Turner Entertainment demonstrated on-air use in late 2005 for broadcast applications.


All of these media are targeted at one metric, the minimum cost per recorded byte, which of course trades off against other parameters. For instance, the cost of the media must be low, but if the cost of the recording device is extraordinarily high, the equation may not work. They all have in common moving media, motors and sensitive mechanical processes, which must be maintained.

Another approach is to use nonvolatile memory for the recording media. In this instance, the transport becomes exceedingly cheap, but currently the cost per megabit of storage is much higher. It does, however, offer other advantages, including blindingly fast random access and no moving parts.

If memory prices fall enough in the future, online recording could become the province of memory devices, and moving media may become the archive attached to them for long-term purposes. Formalizing such a rigid hierarchical storage environment could allow optimization of each portion of the recording process, with online, near-line and removable media off-line storage matching the cost and features required. All we need is 64GB SD cards, which might be in your camera sooner than you would think.

John Luff is the senior vice president of business development for AZCAR.

Send questions and comments