What is all the excitement over the new "Network Computer"? By now, you’ve heard the sound and fury in the press regarding the competing initiatives by Microsoft on the one hand and Sun Microsystems on the other. Both are jumping on the "thin client" bandwagon. Exactly what are "thin clients" you may ask? Well, it seems no one is quite sure. Or rather, everyone has their own idea but the ideas don’t match. The only obvious consensus is that the current PC model costs way, way too much to administer.
Wait a minute! Isn’t the PC supposed to be the current low cost Mecca of MIS? Out with the old mainframes and the rigid central computers and in with the fast, flexible, low cost PC stile computing paradigm has been the battle cry of the hardware vendors for nearly a decade now. We’ve been told over and over again that if you don’t have a new, fast, windows 95 PC on every desktop, you must living in the dark ages. Not only is your company going to be non-competitive, but you will have trouble getting dates. This message has been drummed into us with such force and vigor that it was corporate suicide to even think about questioning it!
But wait, something strange is happening out here in the real world. When you go looking for the bottom line gains, most companies are discovering that many of the benefits and productivity improvements we have been promised are simply not materializing. We have been disappointed over and over again. It seems that each and every promise turns out to require a bigger CPU, more memory, more disk, and more network bandwidth than we expected. Every time we upgrade to take advantage of the newest wizbang mousetrap, something else causes a capacity problem and we find we need to upgrade the next piece of our infrastructure.
The estimated annual operating cost of each and every PC in your shop is between $8,000 and $12,000. Please note that’s not the PC’s purchase price – its the annual operating cost!!! Whoa – exactly when did you sign up for this level of expense? Does anyone know the annual operating costs for a simple terminal? My guess is $250 or less! I’m willing to bet if you had known up front of the actual total costs associated with maintaining your ocean of PC’s, there would be a whole lot fewer PC’s in the world today!
Well, with companies waking up to these costs, there is clearly a lot of pressure to reduce the cost of the PC revolution in order to match it to the new more realistic (read lower) expectations of gain. Suddenly everyone is beginning to talk about measuring the TCO or total cost of ownership. The "thin client" concept is an attempt to reduce the skyrocketing cost of administration for the desktop. The "NetPC" and "Network computers" plan to reduce administration by "dumbing down" the desktop. Lock the cases to avoid weird hardware configurations. Cut out the hard disks and the floppy. Kept the applications and the data on the servers, not on the PC. These and other strategies are attempts to control the exploding costs of support.
Will they succeed? I don’t have much hope, but I’m an obviously biased Pickie. For an independent (someone who probably hasn’t even heard of Pick) look, take a peek at an editorial in the 11/18/96 PCWeek titled "Think thin clients? Do the math first." It points out that at most, 20 to 30 percent of the costs are centrally controllable. The rest are hard costs, local inefficiencies and lost productivity.
Lets take a moment to step back and consider what is happening here. The industry is being forced to make the PC look more and more like a dumb terminal to get it to a point where we can control it. We are being urged to move the applications and data to central servers to back them up and to control them. Excuse me for the moment of Deja Vu, but doesn’t this seem just like a central data base and dumb terminals?? The industry is running around a mulberry bush. We are pushed to run faster and faster so we don’t look up an realize we are going in circles.
Oh, of course the PC monitor is capable of pretty color graphics, and I’m not for a moment recommending we go back to ASCII terminals. Nor am I equating the open inter-networked WWW model to a single system, but the world is surely shifting the recommended system architecture back toward the classic tried and true Pick model, and quickly at that.
Make no mistakes, Pick is far from perfect. The Pick based vendors all need to continue to increase their openness. Most have several good initiatives in this area, and as users our options increase every day. The system vendors also need to continue to create system designs that can work in a networked, client server type environment. Pick applications generally don’t work well in this area yet, but then when you are honest, neither do anyone else’s. Good solutions in the open network world will require new and better Pick capabilities and some user application changes. It is clear to me the winners in this race will be the ones who stick closest to the traditional Pick model and successfully hide the majority of the administrative burdens from the users.
Holy cow Batman, could it be? Could Pick have had the most cost effective model right from the start? Yes, Pick did! The gestalt of Pick has always been uncomplicated solutions – highly useful applications with the absolute minimum of architectural intrusiveness and administrative overhead. This is exactly what these new industry initiatives are striving for and not yet achieving. Think about it for a moment and then drop me a line at News and Review and let me know your thoughts.