<under construction>
Back in 1978 I worked as a field engineer for Nixdorf computer. I worked at our depot in San Mateo California, the San Francisco area. As a depot, we maintained an inventory of spare parts and also provided in-house component repair. We were able to quickly repair a down customer system from the spare parts inventory we had in-house. On rare occasions we had to order out for parts, but at least 95% of our parts needs were serviced locally. In those days an engineer was trained to troubleshoot to the chip. That meant our engineers would use test equipment, such as a logic analyzer or oscilliscope to trace the signals to the chip or part that was failing. Today that skill set is a lost art. It's often cheaper and faster to just replace failing components, such as a motherboard, or a disk drive. Yes, I used to replace crashed heads in a disk drive along with platters. We'd keep a supply of spare disk heads and platters in the office. Of course doing on-site repairs like this meant there would be system down time. It's faster to do FRU (field replacable units) replacement rather than component repair. The tough part is figuring out how many and what kind of FRU's you should keep an inventory of on-site. The other issue is that your data center systems engineers need to be trained in FRU replacement.