Servlets, scanners and a system nobody wanted
In the late 1990s, a couple of us built a warehouse management system for a customer in the United States. It was one of the first web-based systems of its kind. That sounds ordinary now; at the time it meant working with early releases of Sun Microsystems' new Java technologies, servlets, JavaServer Pages and the first application servers, before most of the rough edges had been sanded off.
The design was simple and, for its day, bold. The server ran at the customer's corporate headquarters. Workers in the warehouse carried Symbol barcode scanners. Every receipt, move and shipment went straight into one central system that anyone with a browser could look at.
It worked
From a technical point of view, the project was a success. The scanners talked to the server, the inventory stayed accurate, and the web pages showed what was in the building right now, not what was there at the last count.
Nobody liked it
The people who had to use it were less enthusiastic. It turned out that an accurate, up-to-the-minute inventory did not fit their plans very well. A system that shows exactly what is on the shelves also shows exactly what is not, and when, and who touched it last.
That was my first real lesson about this kind of software: the hard part is rarely the technology. A tracking system changes who knows what, and some people prefer the old fog. Any system that makes operations visible needs someone at the top who wants them visible.
The code, though, was good, and it was about to find a home where visibility was exactly the point.
Next: Walking the rows at the port.
Comments