COTS: Is it Just a for Your Program? - commercial off-the-shelf

Program Manager, March, 2000 by Luke Campbell

The lesson learned was perspective. Our military needs were not only well in advance of our commercial needs, but more disturbing, were also in advance of industry understanding of the concept of speed and performance. The problem was that our market is quite small -- insignificant -- in fact. Military systems are an oddity to industry -- a speck of dust.

We did not give up on commercial databases; we discussed performance with the vendors at some length. After discussions with industry about speeding up their databases, our solution was to go back to the "do-it-yourself" database, and give up on COTS this time. Another lesson learned was that decent system engineering and analysis of industry products is necessary. So called vaporware, or software that is seemingly never delivered, is rampant. Our solution was to take a six-month loss in schedule.

More Vaporware

We also wanted a multi-level-secure environment. After performing surveys for capability and market share, we chose Digital Equipment Corporation's (DEC) Multi-Level [Security.sup. ] ([MLS.sup. ]) system. This lasted for four years before DEC announced that the market for this product was not nearly as strong as envisioned, and the product would be discontinued. The lesson learned was when you use COTS, be prepared for change. Industry moves to the beat of quarterly profit -- period. Fortunately for us, we were not entirely unprepared for this eventuality, and our solution was to fall back on plain-old UNIX, and use our well-designed software architecture for the security features we need.

COTS Computer Performance -- Some Perspectives

The performance growth curve for military computer systems has been virtually flat for the past 10 years because of COTS use. We are only now climbing the curve again. Blasphemy? Let's look at the data. Certainly, there is no argument that the raw power of hardware is light years faster than it was 10 years ago, and there are no 640K memory barriers. But consider the system. Think about the desktop applications you run today and the performance of those applications 10 years ago.

On the negative side, your disk drive is still 90 percent full -- except that today it's 2 gigabytes, while it was 20 megabytes back then. True, you didn't have 100 megabytes of "essential" pictures from the World Wide Web. Or consider word processing. The file size of a page of text -- just plain text -- is 30K, compared to 2K back then. What about performance? Do you actually see the 366 megahertz speed of the latest Pentium compared to the 2 megahertz Z-80? Certainly systems are faster -- but 150 times faster? Efficiency is no longer a part of our vocabulary.

So What, You Ask? Let's Look at the Positive Side

Today, we can easily embed pictures in documents, making them highly readable and understandable. We can ship them around the world at breakneck speeds (assuming the network is up today). We can develop huge spreadsheets for Team Work Plans and Earned Value Management. Who doesn't like having the ability to make a presentation in color, with pictures, sound, animation, and 10 or more fonts using an electronic projector? Is it even possible to still make a presentation with short bullets on a typed sheet, which are copied to a transparency for use on an overhead projector? On another front, new Computer-Aided Design (CAD) software applications speed up designs, basically eliminate paperwork, and perform automatic calculations, saving untold workhours and millions of calculation errors.