Why do computer and computer systems lock up or crash?

Assuming the software is not at fault, or that you don't have conflicts caused either by the improper installation of programs, misbehaving TSR programs, or even port address, interrupt, or DMA conflicts between expansion boards, the most common problem is poor contacts either in the computer (including the expansion boards) or the keyboard, or in the connectors which attach the computer to its keyboard, printer, scanner, LAN's, or other peripherals.

Can these crashes be prevented?

Yes, most of them can be prevented. As good as connectors are, they are still the weakest link in your computer's operation. There are hundreds of these connections in a typical computer system. Outside of the obvious ones such as the connector that connects the computer to the printer, or LAN connections, the computer itself is full of electromechanical contacts.

​

These range from the card edge connectors that are used on the various plugin cards, to the numerous types of sockets into which IC's are plugged. The interdependence of the components within the computer is such that the erratic operation of virtually any one of the contacts in any of these connectors can crash or lock-up the computer. What is even more frightening. if the computer doesn't hang up, vital data or the execution of a program sequence may be altered such that key data is overwritten on a disk and thus lost.

What do these crashes cost?

Many users are not aware of the cumulative costs of computer crashes. But consider: if you're a programmer and have been working on a program for an hour when the computer crashes, and you didn't save anything during that time; the crash just cost you an hour's work not to mention the additional time required to readjust your overview of the program on which you were working. If you're developing a program you'll have to call up the previous version, put your brain in reverse, and try to remember just what it was you did during that hour. ff you've ever had to do this, then you know just how frustrating it can be. Thus, the loss of that hour's work can cost you a lot more than that hour.

Consider a person processing a batch run of invoices. What was the last invoice that was processed? Did the computer program rewrite the FAT file every time an invoice was processed or was it waiting until the completion of the run to update the directory in the computer? Did your program handle all the interdependent files in an identical manner? Do you even know? Well, if it happened to one of your staff and you have to sort out the mess, you're going to find out the hard way!

Consider someone working on a spreadsheet who didn't save it before the computer crashed, not only would it be frustrating to have to go back and figure out just where they were in its development at the previous save, if any, can you be sure that they will be able to match their previous performance?

As you probably know, when a deadline is hanging over your head, it's human nature to try and speed things up by ignoring some of the details which prevent problems, you get "on a roll" and that's when accidents happen. Gee… forgot to save it did you? Now what will the additional delay and a missed deadline cost you?

What about the hidden costs of client dissatisfaction? Could you have lost a number of jobs because you missed a deadline. Sure, you know why it couldn't be met but does your client understand? Then again, perhaps you couldn't really do your best work because of the pressure or because of the lack of time?

What about an engineer involved in a complex CAD-CAM problem? Can you imagine how he feels when part way through a day of intensive work, his input device freezes leaving several hours work in a non- recoverable state?

Or consider newspaper reporters, working against a deadline. How often have we heard of stories which have gone off into limbo because of a terminal crash?

And then, consider the problems imposed by computer crashes in process control or experimental date acquisition. The disruption in even a fail-safe process control could cost more than the computer; and in data acquisition, the whole experiment might have to be restarted from scratch.

​

If you're responsible for servicing a system, wouldn't you like to increase the reliability of the equipment in your charge? And if your part of a team marketing complex systems think of how an increased reputation for reliability could aid you in the sale of replacement or even new systems!

Too often, people tolerate problems on the basis that so long as their system is just as reliable as the other parties. then that's all they have to achieve. In fact, we often run across managers who tell us that on the basis of the proceeding philosophy, "They don't have any connector reliability problems at all!" But when we talk to their service personnel, we find out otherwise.

Connector unreliability is like a time-bomb, waiting to go off in the most unexpected places and at the most unexpected times. Very expensive preventative maintenance programs attempt to stop these unexpected events. But some connector problems elude even the most thorough maintenance, perhaps because of the sheer quantity and diversity of the connections in even a simple system.

What can be done to reduce the number of these crashes?