- Published on
Ask HN: Who's still keeping a DOS machine up because the business depends on it?
- Authors
- Name
- DeepHorus
An Ask HN thread is reviving questions about why organizations still depend on DOS-era and industrial systems after mainstream PC support ended. The post asks whether dBase, Clipper, CLARION, Paradox, CNC controllers, laboratory instruments, and parallel-port devices remain in active use. Readers are invited to share experiences or contact the author through the email in their profile, framing the effort as research rather than a sales pitch. Why it matters: these systems can quietly sustain production, diagnostics, or small-business workflows.
The question taps into a long history of businesses adopting affordable DOS software and hardware for a specific job, then keeping it alive because replacement can be risky or expensive. Database and rapid-application-development tools offered mature file formats, interfaces, and workflows, while industrial PC platforms gave manufacturers direct access to sensors and motion hardware. Such installations often accumulated around the system rather than the software itself.
The thread specifically broadens “old software” to embedded and laboratory machinery controlled through ISA cards, GPIB equipment, and dongles attached to parallel ports. Those categories matter because they combine obsolete operating-system assumptions with timing-sensitive hardware access. Hospitals, factories, repair shops, small offices, and scientific labs may be among the users, although the discussion did not provide a verified census of installations. It is an anecdotal research request, not evidence of a market size.
Emulation could reproduce a DOS environment, but simply booting an image may not preserve an industrial workload. ISA bus I/O, interrupts, direct memory access, driver behavior, and timing can be invisible to a conventional virtual machine. A modern solution might combine a virtualized x86 environment with carefully mapped device emulation, an FPGA or PCIe interface, or a purpose-built controller. None of those approaches is trivial: each requires testing against the original process and a safe fallback.
For businesses, the cost of keeping a legacy machine extends beyond electricity and replacement disks. Operators may rely on undocumented menu paths, specialist technicians, and a single unit of spare hardware, creating concentration risk when the last knowledgeable maintainer retires. Replacing a system can also trigger recertification, downtime, data migration, or changes in regulatory evidence. Preservation therefore has both an economic value and a safety dimension.
Community reaction to the post is likely to split between preservationists who value hard-won hardware knowledge and engineers who see the same machines as a cyber and reliability liability. The request for real-world examples signals demand for evidence, not merely nostalgia. Operators may be reluctant to disclose a facility’s dependencies, while hobbyists and industrial technicians can offer repair patterns, failure modes, and replacement clues. That exchange could shape a viable project.
For vendors, the discussion highlights a narrow but important modernization gap: general-purpose desktop replacement tools do not automatically cover embedded controllers, proprietary instrument firmware, or legacy data formats. A successful preservation effort would need partnerships with machine builders, museums, universities, and industrial-support firms, plus documentation of each device’s behavior. The likely customers are organizations that value uptime more than a polished new user interface.
Looking ahead, the next step is likely a series of case studies rather than an immediate mass-market product. The author says the goal is research around keeping these systems running on modern hardware; responses could reveal whether virtualization, hardware adapters, or restoration of original machines is most practical. If prototypes appear, they should publish supported models, compatibility limits, and safe testing methods. The broader outcome may be a more durable map of digital industrial obsolescence.
