unreliable power
My first experience with unreliable power was at a small office move where they added a few outlets to accommodate the Novell server, phone system, and AT&T Unix voicemail server and associated network equipment. I was trying to keep related equipment on related circuits so a circuit outage would only affect one service instead of everything. It was not quite possible to do it that cleanly, the network switch had to be on the telecom circuit. I was seeing connectivity issues all over. Troubleshooting was a bear. I also had new Cat3 drops to suspect. I for some reason checked the grounds between the circuits for potential. The multimeter showed about 50VAC between the telecom circuit and the computer network circuit. The electrician confirmed and corrected the grounding issue, apparently there were multiple grounds for the building but they were not bonded. Connection issues disappeared.
At the same place one of the user's monitors (CRT) had an annoying "waviness" which was also suspected of causing headaches. When I saw the "waviness", I could see the possibility of headaches. I found the building's electrical feed was directly opposite the wall of the problem monitor. I first suggested rearranging the office to move the monitor, which was declined. I then attempted to shield with a large steel panel both in grounded and ungrounded states. The grounded steel panel diminished the waviness greatly but not completely. They decided to rearrange.
At another job there was a "computer lab" setup in a former cubicle farm space. Fifty or so stations were setup in a very long and narrow space with a server located halfway. This location had better than average policies which included periodically testing the building UPS systems. On the first test after full build out of the lab, half the lab went down, including the server. I had to inform project managers, facilities dept. and others that half the lab was not on the building UPS. The facilities dept. had to admit they knew that but the cost to add was prohibitive. The consensus decision driven from higher up was not to fix the issue. I moved the server to the half that was on the UPS.
At that location I was once volunteered as the escort for the UPS tech for a repair of one of the units. I asked what he was fixing. He replied, replacing the positive battery buss. I walked back across the room to the doorway. I said I wanted to be able to summon help in case of a short circuit. The battery cabinet footprint was about 3ft by 15ft, probably enough energy to blow me through the doorway and he would have been beyond help.
Biting the hand that feeds IT
