[continued] All this tech talk is nice, but to really appreciate the practical implications of this patent, you have to understand the unique position of Ampex's data recorders in the military/aerospace markets. There are two data recorders of choice in wide use by the defense and aerospace establishment: the very expensive DIR1000 (based on Sony's DD-1 format) and the cheaper DCRsi (based on Ampex's DD-2 format). The first DCR data recorders were rolled out in sometime in 1985-1986 when Ampex was still a unit of Allied Signal and when the defense boom of the 80s was in full swing. I recall reading somewhere that these data recorders continue to carry export restrictions. The DCRsi is already CIGSS (Common Imagery Ground/Surface System) compliant and it is a proposed NATO standard.
Thus, Ampex has a large installed base of these purposely overengineered, overbuilt and very expensive ($200,000+) data recorders. The business opportunity for Ampex lies in at least 2 ways:
1) The interface allows Ampex's installed base to extend the useful life of their DCRsi data recorders. At a time when military budgets all over the world are declining in real terms, this is a very attractive proposition.
2) The interface allows Ampex to mix and match their interface with a line of DCRsi data recorders that ranges from the DCRsi 75 (9.4 MB/sec) to the in-the-works DCRsi480 (60 MB/sec) to appeal to a broader market. By increasing their already large installed base (relative to the maximum size of this highly specialized market), they amplify a trend that shows a growing interest by their DCRsi customers in the DIS and even DST drives and libraries. You saw a similar thing happend when the DCT led to additional DIS/DST sales.
The thing to remember about Ampex is that its R&D has always been centered around a high performance systems approach as opposed to a specialized component approach, which is probably why it has developed a wide technology base over the last 52 years. I am curious as to what its plans are in this area. In many respects, semiconductor memory is the ideal storage product: it has no latency, no seek times, no moving parts, and the slowest DRAM is still hundreds of times faster than the fastest disk drives. It is still very expensive, though, and difficult to scale to disk drive dimensions, especially with the 60% annual growth in disk drive capacity.
Last year, however, Samsung and Mitsubishi both indicated plans to start pilot production of 1 gigabit (125 Megabyte) memory chips around 2000. NEC also disclosed plans to start pilot production of 4 gigabit (500 Megabyte) memory chips in the 2000-2002 time frame. These conceptual efforts represent a quantum leap over the current memory products that range from 16-64 megabits or 2-8 Megabyte memory chips.
Given those trends, it is conceivable that solid state memory disk subsystems will be required in future networks especially if you consider that Intel, as the judge, jury and enforcer of Moore's Law (transistor density doubles every 18 months), expects that in 3-5 years, processing power will increase 10-fold! Already, solid state subsystems from companies like Zitel are increasingly capturing the cache space between powerful servers/mainframes and the typical RAID disk drive array at some commercial sites. The cache serves to insure that the processors are fed data at a consistently high rate.
Ampex has disclosed that it has a group of about 200 U.S. and foreign patents pending. Keeping in mind the tight focus that Bramson imposed, these patents provide glimpses into the choices this company is making in a fast moving technology arena. Imagine, for example, the possibilities that exist in its bare bones DCT business (around $10 million in revenues in 1996) if Ampex can combine its work with semiconductor memory and compression with its tape drives, editors and switchers. The ultimate non-linear editor? The ultimate switcher? The ultimate DVE (digital video effect)? Not too farfetched especially if you consider that Ampex developed the first linear editor, the first switcher, and the first DVE box (ADO or Ampex Digital Optics). |
