Even experts make mistakes
Posted by Trixter on June 9, 2008
I’ve been working with classic personal computers for 25+ years. I know all of the precautions in working with older PCs, and yet even I make a mistake so idiotic it just hurts to think about it. I will tell a cautionary tale; see if you can guess the ending before I get there.
At my workplace recently, we were clearing out a 20-yr-old “mini-datacenter” at work (one UPS, one cooling unit, about 50 servers) after a UPS power failure, and one of the machines in the corner was displaying an error message. I hadn’t noticed it before because its screen was usually blanked, but moving over to that side of the room it turned out to be an AT&T PC 6300 WGS. I’m a bit fond of AT&T 63xx machines, so I went over to investigate.
It turns out that the 6300 WGS was originally installed when the datacenter was built in 1987, and its sole purpose was to monitor the UPS. Month after month, the 6300 served as a graphical display of the UPS, with little color-coded pictures of each battery and component, which ones needed service, etc. However, the UPS had failed a few weeks ago (was replaced with power from another source in the building before the servers were affected), so the 6300 had nothing to do now. Some quick research shows that the 6300 WGS model was a much more compatible model; in additional to being able to take a VGA card, it is the fastest 8086 clone I’d ever seen at 10MHz. (I’ve seen faster 8088 clones, but not 8086.) Naturally, I had to have it.
After checking with our company’s obsolete inventory and salvaging procedures, I got the authority to take it home, along with the extremely yellowed VGA monitor, Microsoft bar-of-soap bus mouse, and 9-pin dot-matrix printer. I went over to it and looked at the contents of the hard drive; some directories were boring (Borland Sidekick, PFS:Write), some were moderately interesting stuff (Windows/286, which means the 8086 was replaced with a 10MHz NEC V30 to get it to run), and one very interesting and rare thing (Microsoft Professional Pascal compiler). Rubbing my hands with glee, I ran the program to park the hard drive heads so that the drive wouldn’t get damaged during transport, and powered it off to prepare it for shipping.
Did you see it? Throughout that entire scenario, what did I do wrong? Let’s review some points:
- “was originally installed when the datacenter was built in 1987”
- “its sole purpose was to monitor the UPS”
- “powered it off to prepare it for shipping”
In a nutshell: A desktop computer with a hard drive was installed in 1987 and powered up. All it did was review data coming from a UPS. The desktop computer itself was plugged into the UPS, ergo it had never been powered down. It had been running continuously for over 20 years.
Which means the hard drive had been spinning continuously for 20 years. And I powered it off.
Checking the machine today, to do one last check on it before it was moved, the hard drive first returned garbage, then on a second boot refused to spin up. Which, after a moment of confusion, didn’t surprise me. Oh well; maybe a miracle will occur and I’ll have a second chance of getting data off of the drive, but I doubt it.
My friends, if you want to keep a hard drive running for 20 years, keep it spinning.