Today’s power outage is a perfectly timed example of the kind of event we must be protected against — and currently are not.
When the server is switched off suddenly during important operations — such as saving data, running domain services, or handling disk writes — it doesn’t simply “pause”. It can leave files, databases, and even the virtual machine structures themselves only half-written, which leads to corruption. In the worst cases, the system won’t boot again at all.
This could easily be the reason I’ve been repeatedly battling inexplicable issues while building the system. It takes only a fraction-of-a-second interruption — the sort that often goes unnoticed — to destroy something silently in the background. That suspicion is growing stronger now that this has happened again.
This is now the third time I’ve had to rebuild the server in just a few weeks. It’s becoming very tempting to wonder why.
And this outage didn’t happen while the system was idle — I was right in the middle of critical configuration tasks. And now that work is gone or compromised.
-–
⚠️ This Is Not Just a Technical Problem
The server runs our domain controller, business-critical software, and handles authentication, network coordination, and storage. When it’s knocked out mid-operation:
- Data loss or corruption is likely
- Time and work are lost
- Business operations are interrupted
- Recovery may take hours or days, if it’s possible at all
These risks are avoidable with one simple measure.
-–
🛡️ The Role of a UPS (Uninterruptible Power Supply)
A UPS doesn’t need to keep the server running all day — just long enough for it to shut down gracefully and safely.
- It detects the power loss instantly and gives the server a few minutes to save data and shut down cleanly.
- It prevents crashes, corruption, or damage to virtual disks and Windows.
- It can protect both the server and the software it runs, including our core business applications.
The One S business operations software is no different — it won’t like being interrupted mid-write either.
Why such vulnerability when desktop pcs are ok?
Unlike servers running virtualisation platforms or complex domain services, ordinary Windows 10 or 11 systems are far less vulnerable to power outages. They’re designed for everyday use on laptops and desktops, which frequently sleep, hibernate, or shut down abruptly. Most of their critical filesystems are journaled (NTFS), and the system is optimised to tolerate the occasional improper shutdown — often recovering automatically on reboot. They’re typically running just one operating system with minimal services, no virtual machines, and no background processes writing to multiple disk images or coordinating across a network. By contrast, a Virtual server with live virtual machines and domain controllers has many interdependent services and disk volumes at risk — and a power cut during write operations can easily lead to system-level corruption, VM disk failure, or domain database loss - or all of these .
âś… A Clear and urgent Need
We must install a UPS with enough capacity to give us only 5–10 minutes of runtime, a bit more might be sensible as regards operatives being in the middle of say an order or such like.. That’s all we need to shut down the host system and its VMs in a controlled way. How many devices beyond the server to support is a different discussion.
It will save us from this kind of situation happening again — and costs far less than even a single rebuild of this server.
I can specify a model if needed. But we can’t keep risking this. Power cuts will happen again. The question is: will we be prepared next time?