CSIRO still running Windows 98, NT


blog Sometimes you can’t help but be amused by the longevity of Microsoft’s operating systems. In an otherwise unrelated article on the organisation’s adoption of Internet Protocol version 6, an article published by ZDNet.com.au yesterday revealed that Australia’s peak scientific research agency was still running some copies of Windows 98 and NT4. John Gibbins, IT security operations team leader at CSIRO, reportedly said:

“We still have Windows 98 and NT4 boxes on our network. I think if we searched hard enough we would find Windows 3.1 boxes that aren’t on the network.”

The issue came up because these old Microsoft operating systems (Windows 98 was released in 1998 and NT4 in 1996) don’t easily support IPv6, unlike most of the rest of the network of the Commonwealth Scientific and Industrial Research Organisation (CSIRO), which typically runs a mix of Windows 7 and Windows XP.

To be honest, I can’t say I’m really surprised — but I am amused. And I bet none of Microsoft’s NT4 developers expected their glorious accomplishment to still be in use in 16 years’ time. Perhaps my biggest question is why these Win98 and NT instances haven’t been virtualised yet; it seems like that would be the normal approach these days, if the platform still needed to be used for some specific software packages.

Image credit: Screenshot of Windows 98 default desktop, sourced from Wikipedia


  1. They probably haven’t been virtualised because nobody is game to touch the systems at all, for fear of breaking them entirely. I know I’d be particularly reticent about so much as restarting a Win 98 box these days.

  2. The reason why they probably haven’t been virtualized is likely due to what they’re doing.

    At my company we have some machines running old operating systems, nothing as old as this, but quite old. The reason we maintain them is due to the software they’re running, and the peripherals they have. For example, we one machine which runs an old 16bit application which is a control/reporting interface which uses some ISA card which connects to the machine using some proprietary protocol. This machine is used for doing… science. (Not sure exactly, but it’s used somewhat regularly)

    To replace the machine would cost the business a lot of money, with very little gain. The machine in question isn’t on our network, and literally does this 1 task.

    We’ve got other machines that need to be connected to specialized instrumentation via serial/etc, some of which we’ve managed to virtualize or upgrade, but others which failed when we tried to.

    Overall, some of this old stuff is hard to get away from. You can’t do much about it, short of replacing the machine, which can run into the millions of dollars.

  3. This would not suprise anyone who’s done IT in a research organisation. Labs around the world have ancient PCs which are key parts of bespoke experimental setups, logging data or controlling apparatus through barely custom built 8-bit ISA controllers and so on.

    It’s not practical or even necessary to replace most of them, and many of these systems don’t even need a network interface. The biggest problems for organisations running these relics are things like sourcing spare parts, finding IT staff who know how to cable up a 5.25 inch FDD and who can make sense of that 20 year old qbasic program that drives the whole thing.

      • There’s also a lot of old industrial hardware. Built in the 80’s/90’s, designed to have lets say a 30-year lifespan without issue…and so naturally the electronics side runs with old floppy as input or a 486 as a control unit. possibly running an early ms-dos, and with ISA output cards that mean you cant just upgrade it.

        I keep working hardware around from the 286 era onwards, just for the rare occasions that they’re suddenly and urgently needed.

        • I think I’ve still got an ISA card around here somewhere, from our first computer….XT8086….My dad jerry-rigged it with 20 MEGAbytes of HDD and 2, count them people, 2 x 5.25″ Floppies AND a 3.25″!

          Wish I’d never given it to my school. Brilliant keepsake….and they dumped it after the comp. teacher who loved it left.

  4. Wait a minute – aren’t all those scientists on the ‘gravy train’? Shouldn’t they be able to afford new PCs?

    That’s what the global warming deniers tell me, anyway… :-P

    But seriously – yeah, probably need to run old software to interface with old hardware… we had a client who was running an NT4 box until very recently, to keep an old piece of software running. Mind you, it wasn’t on the network, so IPv6 wasn’t really an issue. Getting the data off with no USB support was a pain, though (especially when the software produced datasets measured in hundreds of megabytes!).

    • The old serial null modem cable has worked wonders for me in the past to get data of these types of systems.
      Although slow is less painful than shuffling disks.

  5. Doesn’t surprise me at all. As others have said they’ll be running hardware that was designed for these OSes and that haven’t been updated.

    One place I worked at they had a weather station on the roof and the box hooked up to that for the reporting was Win ’95 because that’s what it was originally setup with and what the software ran on and they had no business requirement to upgrade.

    • I have and old old Dell laptop here with no battery(long since shot and haven’t been able to replace but it doesn’t matter) still used with one specific bit of kit because I haven’t been able to get the software to run virtualized or under a newer OS or hardware and still successfully communicate with the bit of hardware it controls.

      Went through a significant number of RS232 interfaces (USB, PCI, PCMCIA ect..) for another bit of hardware when even though the software was perfectly happy on even windows 7 64bit the hardware just wouldn’t talk to any old RS232 port.

      Thankfully all these are control systems and don’t need to be on the network to dump data. Although my stock of floppy disks for loading configuration files for an old DOS based embedded system is running low.

  6. Equipment we have that I would like to bin.

    Overhead projector, still have one and will bin it once the last lamp we have finally gives out. Did get used for some paying work a month or so back.

    Old Panasonic professional SVHS player, – Get fired up and test once a year just in case because we know we will never be able to get another one if we need it in a hurry. Had two and now down to one still working.

    Varies video interfaces that support obscure video connection that haven’t been in use since the early 90s. Only keep these because they don’t take up much room and I want to one day cause a heart attack when I say yes we can support this 20 year old bit of kit at a conference. Although did use one of them recently to connect an old arcade machine to a LCD screen.

    About a year after we retired and sold our fleet of Kodak SAVs(slide projectors) after having them sit on the shelf not being used for 5years we had a request for one.

  7. Doesn’t surprise me.

    When I was at Wollongong Uni, they were doing some pretty complex biorganic chemistry research in their main labs…..on Windows NT4. Mainly this was because the ASM unit they were using was that old. They didn’t have the money to replace it, at about $400 000 a pop and it’s interface was so touchy, anything even REMOTELY related to XP made its’ sphincter close up faster than the airspeed of an unladen African Sparrow.

    They had a BITCH of a time getting the data off. The ASM used to produce files about 25-30mb per run and the computer had had its’ cd-rom replaced about a dozen times. But they had to keep sending them away to get repaired because new ones wouldn’t take properly, or they’d drop half the data because of buffer underrun.

    And that was just their ASM. Their old HPLC machine used a unix system that connected to a Windows 98 machine. It was eventually replaced and we used the new one when I was there, running XP I think. But if there was contention on use, you used to see the Post-Grads digging away for the right floppy disk in the lab….

  8. The joys of highly specialised software and systems.

    Typically, it isn’t so much a lack of wanting to shift systems, as much as whether it’s a) still developed, b) still supported c) upgradable to newer host operating systems and most importantly d) financially viable to BE upgraded.

    You’ll see this in most any specialist large business or, more particularly, research; where systems have a very focused purpose, potentially written for specifically that one task, perhaps controlling a piece of diagnostic or research related hardware.

    Things don’t always change at the speed of light. ;)

  9. Not in the least bit surprising. I know some companies that do research: chances are that those PCs are display units for highly specialized (e.g. one off) hardware, where there were about 5 built in that era, there’s a new model now (but it costs more than a house), and nobody is game to re-write the device driver for NT kernel or whatever…

  10. Most of the issues relates to getting support for proprietary hardware such as ISA cards, LPT licensed dongles or very specialised software where the company no longer exists. None of these place nice with a newer OS, and chances are cost more than the latest computer itself!

  11. Magnificent goods from you, man. I’ve consider your stuff previous to and you are just too excellent. I really like what you have acquired here, certainly like what you are saying and the way during which you say it. You make it enjoyable and you still take care of to keep it wise. I cant wait to read far more from you. This is really a tremendous web site.

Comments are closed.