Since MRP was invented about 30 years ago, a lot of progress has been made in both software efficiency and hardware capacity. The rule of thumb in the beginning of MRP application was to run the regenerative net-change program and calculation every weekend. That was because the software program and corresponding calculations took hours to run, and often nothing else could be processed while the room-size computer churned the data, layer by layer. Even more interesting was the philosophy that running it too often introduced a large amount of unnecessary system noise (variation) into the manufacturing process. It is actually quite humorous to think about that today. That system noise was driven by changes in customer demand, and the reason it was called noise was because the manufacturing managers did not want to change over machinery setups in the shop. They wanted to run long lot orders. If MRP was not run very often, changes still happened but they were not communicated to the shop, which artificially created stability in the schedule. This luxury was offset by huge amounts of inventory in finished goods to avoid this noise. Expediting was often funded willingly, and heroes were created in the shop and in purchasing daily as companies like FedEx became household words. Today, things are very different, from the scheduling philosophy standpoint.

Customer demand is why organizations are in business. Ignoring it is foolish and costly. If the existing production process is not matched to the...