Mission-critical IT systems don’t always need the latest tech
Infrastructure Management

Mission-critical IT systems don’t always need the latest tech

Earlier this year there was a fuss amongst IT commentators when it was shown that the British Royal Navy's new warship, HMS Queen Elizabeth, was apparently running Windows XP. Scorn was poured upon the ship's designers and engineers, and parallels were drawn with the USS Yorktown, which had to be towed back to port after its critical systems running Windows NT failed.

Amid the noise, there was some speculation that the Windows XP image was just a cheeky spoof to fool journalists. Perhaps it was actually a screen-saver for Linux... or Windows 3.11. Regardless, the Ministry of Defence stated that XP would not be used when the ship is operational.

But does this matter? Is the age of the software or hardware important in such projects? To answer that question, some context is required.

Despite the best of intentions (see the first resolution in this article), I've failed to stop collecting vintage computers. But this affliction means I get to see a section of the IT industry that most analysts and journalists don't. It's a section that's not new, trendy or ground-breaking, yet it demonstrates the truth of the old saw: if it ain't broke, don't fix it.

For example, I like the soft amber glow of Hercules/MDA CRT monitors [main image]. They're easy on the eyes, don't cause problems with sleep cycle disruption and are well suited to draft writing. They're also about 30 years old, which means they tend to develop faults. Some of those are repairable, others aren't.

The vintage computing community isn't much interested in these displays, preferring CGA or VGA. So you'd think I'd be able to find replacement monitors cheaply on eBay. But no: any that appear are quickly snapped up, sometimes with prices in the hundreds of dollars/pounds/euros range. Why? Because there are still thousands of industrial CNC machines out there that use Hercules/MDA monitors coupled to ancient computers.

When those monitors fail, it makes sense to replace just that component, not the entire system. The price is cheap compared to the tens or hundreds of thousands for an entirely new CNC machine. So the 30-year-old IT systems labour on, booting up from ancient MFM hard drives or 8-inch floppy drives that, despite the layers of dust, still do the job they were designed to do.

An entire ecosystem has sprung up around keeping old industrial technology going. There are converters to adapt CNC machines' video output to newer TFT screens, sometimes combined in a single unit. Then there are the Gotek drives that emulate a 3.5-inch floppy drive using USB memory sticks. Die-hard Commodore Amiga fans may love them, but they were originally intended for industrial knitting machines and still sell into that market by the thousand. Again, why replace an entire machine just because one component is failing?

As some recent threads on the Vintage Computing Federation's forum show, there's a lot of old technology still working quietly away in critical systems. From here:

"I service several 8" floppy drives each year, and occasionally sell spare drives to the broadcast studios to keep their old editing machines running. They tell me that replacement machines would cost them $250K-$500K each, and the old machines still work fine, so why replace them."

"Groundbreaking [of a nuclear power station] was in – hold your breath – 1973. Consider what upgrading the computer process control would entail with endless review, regulatory approval, licensing…."

And from here:

"You don't throw away a $50k+ network analyser just because it has a floppy drive!"

Is a nuclear power station running pdp-8 systems more or less vulnerable than if it were running PCs with Windows 10? Is a stand-alone medical scanner more or less useful than one that can easily be compromised by malware?

These aren't easy questions to answer. Some major engineering projects have construction schedules of 20 years or more. That means the ones coming into effect today were designed up to 20 years ago. It's not always practical to change the design now, because that would mean re-testing and re-evaluation of the entire project.

So what are the IT options when facing such major engineering challenges?

  1. Code from scratch. Hire embedded systems programmers to develop a new operating system and bespoke applications from the ground up, designed specifically for the task in mind. This has the advantage of code simplicity: you wouldn't include any code that you don't need. But reinventing the wheel doesn't mean you'll make a better wheel. Any such system will have bugs and vulnerabilities of its own. Obscurity might hide those for a while but eventually they will be found, with potentially devastating effects.
  2. Use a commercially-available operating system, such as whatever version of Windows is current. This would include professional support for fixing errors and – at least in theory – a version with just the core, essential components you need.
  3. Use an open-source operating system, such as Linux, and software. Similar to point two except that the support would come from a community and the code would be open to scrutiny – though that doesn't guarantee that it's safe.
  4. Update whenever something new and shiny comes along. As appealing as this might sound to armchair analysts, installing new systems just because they're new doesn't make sense from an engineering perspective. It introduces a wide range of new scenarios that must each be painstakingly tested, often for little or no real gain.

The flight systems in the Space Shuttle, the most successful crewed orbital launch vehicle in history, ran on ancient 8086 processors right up to its retirement. By the end of the Shuttle's serviceable life that computer chip was an antique. Should it have been replaced with something newer just because that was possible, perhaps a Pentium chip… that later turned out to have the FDIV bug?

There are some circumstances in which it makes sense to continually upgrade hardware and software, such as for employees' workstations. There are other scenarios in which any upgrade increases risk rather than decreasing it. The trick for large organisations – and for analysts – is to understand the difference.


«Why is the first coding bootcamp closing?


Should CIOs take employees offline to improve security?»
Alex Cruickshank

Alex Cruickshank has been writing about technology and business since 1994. He has lived in various far-flung places around the world and is now based in Berlin.  

Our Case Studies

IDG Connect delivers full creative solutions to meet all your demand generatlon needs. These cover the full scope of options, from customized content and lead delivery through to fully integrated campaigns.


Our Marketing Research

Our in-house analyst and editorial team create a range of insights for the global marketing community. These look at IT buying preferences, the latest soclal media trends and other zeitgeist topics.



Should the government regulate Artificial Intelligence?