When Simon Sharwood revealed that an Aussie operator has just retired a server that been running flawlesssly since 1997, we figured it would prompt a slew of one-upmanship comments, and we were right.
What we weren’t quite prepared for were the diversions that one aging server prompted into the nature of time, space travel, and fans.
Yes fans. Because that was the first topic that came up in comments.
Voland’s right hand was first to the punch, and not at all surprised about the box’s longevity.
While I am not surprised about the electronics and the disk I am still surprised about the fan MTBF. So the 64000$ question is who made the fans - I cannot think of a single fan vendor from ~ 1997 which would deliver a fan with MTBF of > 5 years (even with a "speed reduction" resistor).
Yes, they made fans right in them days. So much so, that Jess has found other uses for vintage PC fannery.
In about 2002 (+/- a year) I set up a large fan from an old olivetti 386 to act as an extractor in the entrance to my house.) It was running 24/7 (powered by an old gamegear psu). It packed up about 9 months ago. And that was in a harsh environment, dust, damp and temperature extremes.
Still, things moved back on track fairly quickly. And taking an entirely unscientific approach here, it seems that when it comes to putting in the years, Sun kit takes some beating.
One anonymous coward, recounted:
we recently permanently retired a couple of ancient dust-covered Sun Netras, which had never suffered any sort of hardware failure apart from a couple of SCSI drives which had given up. We had originally sourced all of that old kit second hand, including the SCSI drives, so no suprise about the disc failures.
The retirement was only necessary because we're no longer running the service they support. The boxes themselves are still perfectly happy, and in a way I'm sad to see them go...Our longest continuous uptime was 3.5 years for one of the Netras, running Sparc FreeBSD ... and the downtime was only because someone managed to unplug our box by mistake.
And gold star Chris King added:
Before I forget, the previous holder of the "Father of the Network" title was a Sun 3, which got the "I AM 20" badge before finally being replaced by a Linux workstation.
As calmeilles noted, “there's little predicting which will and which won't [last].”
I had a Sparc 20 that was in use (albeit not always for the same purpose) from 1995 until 2009. When finally closed down for decommissioning it showed 1953 days uptime.
Sun didn’t have it entirely its own way. There are Wintel boxes that would make your grandparents go misty-eyed still chugging away out there.
NT3.51 would run for eternity and never need a reboot. NT4 was a little buggier due to more stuff allowed into ring0, but those were the days when a windows os could just go and go and go. Anybody still running os/2 applications? That was also extremely stable.
So, yes, they really made things to last back in the 90s, didn’t. What’s that you say? What about pre-90s kit? Robert Helpmann threw things into stark perspective, pointing out that:
It IS Rocket Science Not something I had the joy of working with, but I believe the computers on the Voyager spacecraft (see also) have to be at least in the top 10 in this respect. No-one's been round for a service call since 1977.
There. It happened. Somone mentioned space, so everyone else has to show off their physics chops - in this using the original server’s Seagate hard-drive as a launchpad.
Does that mean that, due to relativity, there is a tangible time difference between the inside and outside of the disc?
To which Richard Cox helpfully replied:
Yes, since any velocity or space-time distortion will give a change. But I doubt it will be much. Assuming the outer edge of the data area of the platters is 3", I get a linear speed of 21.6m/s. Which gives a adjustment, of special relativity, of 0.0026%. However, this is non-linear motion so general relativity applies. Which reverses the effect. But I've no idea by how much.
When toughluck pointed out that some rough calculations mean the disk travelled: 25.6 million miles at the rim or 6.6 million miles at the spindle
When you consider that that's barely more than a quarter of the way to the Sun, that's actually a bit disappointing :)
And if that doesn’t bring you back to earth, how about a little DEC fest? You don’t remember Dec? What are you, under 40 or something?
We had a VaxCluster at work that ran from around 1995 to 2011 so that's 16 years, was only turned off about 2 or 3 days during that time for power supply work - https://en.wikipedia.org/wiki/VAX_4000 In it's glory days it supported about 1000 users, in it's last days was just used for legacy COBOL print jobs to a huge old Digital printer - http://www.purplewave.com/cgi-bin/mnlist.cgi?130409/W9416
Chris King topped that, adding:
That's not unusual for a VAX. We had a 4000 series that was installed in 1992 and was finally decommissioned in 2013 - no doubt there are older systems out there still providing service.
But let’s face, we know those things were built to last. It’s the stuff that wasn’t, but survived against the odds that really tickles our fancy.
Nigel 11 worrying told us:
I have to hand a Compaq 80286 system that still runs Windows 3.1 perfectly on its original hardware. To be fair, it's not a server, and has spent a good fraction of its life switched off. It was the control system for an expensive piece of lab apparatus that was (relatively) recently scrapped because it broke beyond our in-house abilities to mend it, and the manufacturer went bust many years ago. I don't have the heart to throw it away. Will probably fire it up annually at Xmas party time until it won't boot, or until I pop my clogs, or until the National Museum of Computing or NASA wants it. It has a 5.25 inch floppy drive.
Mmmm, it’s actually worrying how many “labs” are running kit that’s older than the average PhD candidate, these days.
As Iafnlab points out:
While not as old as your 286, I support a white box Windows 98 PC hooked up to some brand-name lab equipment. The lab equipment maker is still around but no longer supports this particular piece of equipment. The PC itself is pretty archaic. It doesn't have USB or networking capability, so when the researchers want to send data to colleagues, I have to pull the hard drive and hook it up as a D: drive in an old Optiplex that uses PATA connections. From the Optiplex, I move it to the lab folder on the server, and they can send it wherever they need to. The PC is only used on occasion and spends most of its time turned off.
And should we be worried about BebopWeBop's dayjob?
Well I have been responsible for porting instrumentation and analytics software for a 25 year old Sun installation (it just kept on working and reporting and no-one really noticed it) at a radiological monitoring service. And I am pretty sure that is not the oldest thing lurking there - given the number of PS1 stations attached to instruments I have seen while tracing physical lines....
Talking of telecoms, Teamevil warns:
We've got a 486 running SELTEK Voice Manager, connected to an equally stable Avaya Index Switch that's been running since 1997. It's only down time, power cuts.
486? Fantastic. But let’s go back a couple of generations. Inspector71, what have you got to add?
We have an old Compaq 286 from I think about 1988 still running an important piece of test gear. 1MB RAM,40MB hard drive, DOS 3.0. Runs for about 12-13 hours a day every week day. It has a custom ISA board and and custom software. We are only now getting around to looking at a replacement. The only backup we had was a fan fold printout of the software. I managed to get the files off it with the help of a floppy based FTP server. It is built like a tank though.
Yes, the '80s, that what we want to hear about. Wolfetone is clearly an aficionado.
I own a IBM PS/2 Model 70 (built circa 1987) which runs Windows 3.11 that I bought about 3 or 4 years ago from some bloke in London. It was the same model I had as my first PC in 1997, so I made the trip from Birmingham one evening to get it. That, I'm happy to say, still works. So yeah it would still be working now if you kept it. Along with that, I have an IBM PC-XT 286 which was built in 1984 if my memory is correct, running Windows 3.0. Still works like a dream, and it's a lovely thing to hear it's IBM Hard Disk whirring down after being switched off. These two machines are only used the odd time, sparingly, though. Neither are servers, but I think they stand as a testament to what IBM used to build and the quality of their machines. Lenovo should take note of these.
But Adam Hartfield deserves a special mention. Not just for having creakingly pre-historic kit, but actually writing his comment from there.
The IBM Model M I'm writing this on was made in April 1988. I think I started using it in the late 90s when it came from another office that we closed. They'll have to pry it out of my cold, dead hands.
So, let’s wrap it up with three highly honourable mentions. It was inevitable that a BBC Micro would make the list, and James Pond was in there at the dawn of time. Or at least the dawn of the '80s.
My BBC B Micro from 1981 is still running, along with the extra 32K Watford Electronics RAM expansion board and Cumana 5.25" disk drives. Can't say it has been in continuous use (from '85 to 90 it was switched off whilst a Tiko 486 took its place) but I still do a bit of BASIC and assembler programming on it from time to time. Sadly my Sony cassette player stopped working but I can still play Elite from floppy.
On the business end of things, Alan Sharkey takes us back to the early early days of luggable computing.
old kit? I've still got an original Compaq luggable (circa 1983) and a Compaq plasma portable (1985?) in the loft. Both running DOS and both still work - but I can't do anything with them because I have no 5.25" floppies to get info on and off (and they have no network, of course).
GrumpyKiwi deserves a mention, purely for heading his comment: The WANG that would not die
In 2014 we finally decommissioned the WANG "server" we'd been running since 1992. Had to be decommissioned despite all the grumblings from the Grognards as the guy who looked after it was retiring and moving to the Gold Coast and there was no-one left anywhere else who knew how to look after one. Can't say that I shed any tears.
And finally, a tip of the hat for Saxicol for his heroic efforts in trying to claim the prize for the oldest kit, and for working at the Crystal Maze in Pembrokeshire. Yes, Pembrokeshire's Crytal Maze was a real thing apparently.
I can't beat 19 years but the games in the Crystal Maze in Pembrokeshire ran for 15 years on 386's with 4MB of RAM and 40MB HDD's. I spent 5 years coaxing then into life every morning. The non-booting ones mostly needed to be left on, warming up for a while, then rebooted again. At the end getting HDDs that those 386's would recognise was getting to be a real problem and when I could get one it was expensive compared to the current GB drives. The OS was MSDOS 5.2. There were 20 in total. So 20 x 15 = 300 years, no?
But our final mention goes to Andy Taylor, not just for having the oldest anyone claimed, but for the sheer chutzpah of asking for a certain hard-to-find add-on.
What about the TAC? At the National Museum of Computing we have one of the pair of TACs (Transistorised Automatic Computer) from Wylfa nuclear power station in Anglesey. These machines were in service between 1963 and 2004. Following a relatively light restoration, which included swapping some circuit boards for unused original spares, our TAC is now working. We are now looking for a nuclear power station to control with it.
So there you have it. Thanks for all your comments. You've made a lot of old men and women very happy, and no doubt left a lot of young ones really quite confused. And if you know of a nuke facility going spare, we know someone in Bletchley who'll take it off your hands. ®