
     The Future of the Workstation 
     From the Monterey Bay Area Users Group 
 
     Recently, I was asked by several students at a local college what I 
     saw as the future of "workstations".  The question was being asked 
     in the context of a practice marketing study for a workstation vendor.  
     I took some time to reflect upon their questions remembering that I 
     had been asked almost the identical questions years back. Only then 
     the subject of the study was mini-computers. 

     When first asked, I had just begun working with microprocessors, 
     Motorola 6800's under the guidance of one of New England's more 
     brilliant systems wizards, Bill Taylor.  He was most fond of saying, 
     "One user, one processor!" --  which he later changed to, "One user, 
     at least one processor".  I was a mainframe and minicomputer 
     programmer, and had difficulty seeing how the cost of a processor 
     could ever be justifiably borne by one user.  These microprocessors, 
     they were wimpy little toys, hardly suitable for anything involving 
     heavy computation! Sure, they could be used as controllers, or in
     cash registers, but they would never take the place of real 
     computers.  How could you do a payroll, calculate the path of a 
     satellite, or engage in large scale data analyses with one of these? 

     The PC now on my desk has 20 times the physical memory of the 
     mainframe I was writing code for several years ago.  It is faster, 
     vastly more interactive and oh so much smaller.  I could heat a very 
     large building with the power required for that old beast; that 
     baby was water-cooled! 
 
     The future of workstations.  

     Workstations are essentially minicomputers made small.  By and large, 
     these systems run some variant of UNIX, a 1960's style operating 
     system made popular by Bell Labs freely giving it to universities
     "for research purposes".  UNIX is a multi-user, multi-tasking 
     operating system perpetuating the philosophy of the minicomputer, 
     "a processor must be shared among users to justify its cost".
     UNIX predates MS-DOS in the marketplace, yet it represents a very 
     small fraction of the desktop market. 

     Windows has a larger market share now and Windows-NT is projected 
     (by Microsoft) to outstrip it even further.  It is ballyhooed by 
     its proponents as the replacement for DOS, yet this has not happened.  
     Its market share is not even growing at the same rate as 
     oft-pronounced dead MS-DOS. Why not?  Simple economics...
 
     Why Does It Cost So Much? 
     
     UNIX programs cost more because there are fewer UNIX machines 
     (smaller market).  One could argue that there are fewer UNIX 
     machines because they cost more.  Furthermore, workstations do not 
     have binary compatibility (a complicated way of saying that an 
     application for one will probably not work on another).  

     The incompatibility problems shared by DOS users pale in comparison 
     to those of workstation software developers. This further divides 
     the market, pushing up costs. As a consequence, it is not uncommon 
     for vendors to charge 4-6 times the price of a DOS application for a 
     UNIX version.  It's often justified saying that the workstation 
     costs more and so it is fitting that the software should as well.  
     This is hogwash.  Daily, I read of applications which can only be 
     brought to fruition on workstation machines. 

     Unfortunately, this is advertising puffery.  PCs have come into 
     their own right.  With powerful multi-tasking operating systems like 
     OS/2 and Windows-NT, the death knell for workstations is beginning
     to toll, as it did for mainframes and minicomputers in past years.  
     Users want the benefits that workstations provide, but aside from 
     multi-tasking and networking which are already available to high-end 
     PCs, what are these benefits? Bigger screens? 486 and 586 
     (er, Pentium) machines share the same disk drives, the same network 
     communications, and the same or better (but certainly cheaper) 
     add-on devices.  PCs consistently demonstrate better, more usable, 
     friendlier and just plain more software than workstations.  PC 
     software gets better faster and in greater abundance than that 
     available for UNIX platforms.  The PC market itself is larger, more 
     aggressive and more compatible than the workstation market is likely 
     to ever become. 
 
     The Toymaker is Winning 

     PC users are moving up, getting bigger, faster and more powerful 
     machines.  Application vendors who previously ignored these "toys" 
     are seeing competition brewing from below and are racing pell mell 
     to "port down" to OS/2 or (more likely) Windows-NT.   Now if you had 
     a program which you could offer to a market 100 times as large, 
     wouldn't you "port down"?  Workstation vendors are betting their 
     future that they can reduce costs and achieve widespread portability 
     faster than PC vendors can push up performance.  The odds on this 
     bet are very, very long. 

     So, sitting here smugly in front of a 486 clone, multitasking like 
     a mad fiend, I must conclude that the future for workstation vendors 
     is looking pretty grim.  As I reflect back upon what Bill said long 
     ago, I am comforted knowing that my computer has at least four 
     processors, each performing separate functions, but all working just 
     for me.  Besides, being an only child, I really never learned (nor 
     enjoyed) sharing. 
