@S1-A.ARPA,@MIT-MC.ARPA:amon@cmu-ri-fas.arpa (06/25/85)
From: Dale.Amon@CMU-RI-FAS I am continuously surprised by comments from the aerospace community that lead one to believe that modularity and upward replacement are a new discovery. Most of us on this net know that is not the case. A remember arguing with a friend in the planetary sciences about the need for plug in bus type satellites some years back, even before the Mariner Mk 4 was brought up as a 'new' concept. Now I'm hearing that NASA has decided to reinvent the concept again for the space station? I am leading back to my comments on the space shuttle computers. IF the shuttle electronics are designed to the standards which are considered NORMAL in the computer industry, it should be possible to pull the box and plug in another standard machine. As an example of this philosophy, I point you to the DEC Q-Bus. Years ago you got a quad hieght board that held a CPU, and another quad height board for 4k of ram. The system has gone through growth and evolution to the point where you can put the equivalent of an 11/70 in one slot and give it a few meg of ram to play with using a couple more slots. And you can run the same OS and programs you ran before, AND use the same peripherals you had before, except that they are upward compatible and have evolved also. I often suspect that the aerospace engineers have been used to such long design cycles that they never have had to face obsolence so immediately and personally as have those of us in the electronics rat race. I personally went through 3 complete product line design cycles in less than 7 years at a previous job as an R&D manager. It took that level of effort to keep our noses even with the competition. You project the curves on RAM density, and target your systems to reach prototype when the vendors are sampling, and to go into production when the first production runs start. When you deal with that kind of cycle, modifiability and upgrading become a way of life and a PRIMARY concern of the design effort, not an interesting abstraction. You may have to change parts in the middle of the stream because of availability problems, pricing, changing market requirements, the mood of the CEO's wife last night... Once you have learned this mode of thinking, anyone who designs otherwise seems extremely amateurish and quite foolish. What's future shock? Hasn't everything ALWAYS been changing every year? Dale Amon