Load Shedding IT Continuity: Keeping Your Business Running

The Damage Happens When the Power Comes Back

Ask anyone who fixed computers through the Stage 6 years what destroyed the most equipment, and the answer is not the outages. It is the restorations. When a substation re-energises, the surge that travels down the line has killed more power supplies, motherboards and hard drives than any blackout ever did. The blackout just switches things off. The switch-on is the assault.

Load shedding itself has eased considerably since 2024, and that is genuinely good news. But nobody responsible for infrastructure believes the problem is solved. Eskom still runs an ageing fleet, shedding still makes brief returns when enough units trip at once, municipal networks are in worse shape than the national grid, and cable theft or a failed substation can take out a suburb for days with no schedule and no warning. Power resilience for your IT is no longer about surviving Stage 6. It is about accepting that South African electricity will stay unreliable in less predictable ways, and building systems that shrug it off.

Why Power Instability Is an IT Problem, Not an Electricity Problem

Most business owners think of an outage as lost hours: the office goes dark, work stops, everyone waits. If that were the whole story, the fix would be a generator and patience. The real damage is quieter and more expensive.

• Dirty power kills hardware slowly: Surges on restoration, voltage sags and brownouts all stress power supplies, capacitors and drives. Equipment that lives through repeated hard power cycles does not fail on the day. It fails three months later, mysteriously, out of warranty.
• Unsynced shutdowns corrupt data: A server or NAS that loses power mid-write can corrupt a database, break a RAID array, or damage the file system itself. The classic victim is the accounting database. Sage or Pastel stops opening, and the backup turns out to be a copy of the already-corrupted file.
• Backups fail silently: Backup jobs run overnight, exactly when unattended outages do their worst. A job that dies at 2am does not send anyone a WhatsApp. You discover it the day you need the restore.
• Networks develop mystery faults: Switches, routers and access points that reboot repeatedly end up in odd states. VoIP calls drop, printers vanish off the network, and everyone blames the software.

UPS Buying: One Box, Two Very Different Jobs

The most common mistake we see is a single UPS bought to do two different jobs, and doing both badly. Before you spend anything, be clear on which role you are buying for.

• The graceful shutdown role: For desktops and servers, the UPS is not there to keep the machine running. It is there to close it down properly. Five to ten minutes of runtime is plenty, but only if the UPS is connected to the machine by USB or network and configured to trigger an automatic shutdown. A UPS that keeps a server alive for ten minutes and then drops it mid-write has achieved nothing except postponing the corruption.
• The runtime role: Your fibre ONT, router, main switch and a couple of access points together draw perhaps 30 to 60 watts. At that load even a modest UPS runs for hours, and the small lithium units built specifically for routers cost under R1,500 and manage four hours or more. That one purchase keeps the whole office online for anyone with a charged laptop.

Two practical notes. Servers and better desktops want pure sine wave output; the cheapest UPS units approximate mains power roughly enough that some modern power supplies refuse to run on it. And never plug a laser printer into a UPS. One print job will flatten it.

Inverters and Batteries: The Trade-offs

For keeping people working, rather than machines alive, inverters make more sense than racks of UPS units. The trade-offs are worth understanding before you buy.

• Transfer time matters: A decent UPS switches to battery in milliseconds. Some inverter trolleys take long enough that a desktop restarts. That is fine for lights and Wi-Fi, and wrong for anything with a hard drive. Check the spec sheet before assuming an inverter can protect a computer.
• Lead-acid is cheap until it is not: Lead-acid batteries handle a few hundred deep cycles. Daily load shedding killed them inside eighteen months, and the replacements often cost more than the original saving.
• Lithium costs more up front and less per cycle: LiFePO4 batteries manage thousands of cycles, discharge deeper safely and charge faster. If your backup power will actually get used regularly, lithium is almost always cheaper over five years.
• Size for essentials only: Laptops, screens, the network rack and some lights. Not the kettle, not the aircon, not the laser printer. Write down the actual load list before you let anyone quote you.
• Solar changes the economics, not the engineering: Panels extend runtime and cut the bill, but the inverter and battery do the resilience work. Get the backup right first and add panels when the budget allows.

Keeping the Internet Alive: LTE and Fixed Wireless Failover

Here is the odd irony of fibre: during most outages the fibre network itself stays live, because it is powered somewhere else. Your connection dies anyway, because the ONT and router on your wall have no electricity. A small UPS on the rack solves that. But sometimes the local node does go down, or a contractor with a digger finds the cable, and then no amount of battery helps.

The answer is a failover router with an LTE or 5G SIM that switches over automatically when the fibre drops. Business-grade routers do this out of the box, and the ongoing cost is a data SIM you hope never to use heavily. Two refinements from hard experience: put the failover SIM on a different mobile network from the one your staff phones use, and remember that towers run on batteries too. During long or repeated outages tower batteries deplete, and in some areas they get stolen, so mobile coverage degrades exactly when you need it most. Two independent paths beat one, and fixed wireless from a local provider makes a solid third.

Cloud-First Is Load Shedding Insurance

The businesses that sailed through the worst load shedding years were not the ones with the biggest generators. They were the ones whose systems did not live in the building. If your email, files and core applications run in Microsoft 365 or Google Workspace, they run in data centres with redundancy measured in days, not minutes. A charged laptop and a phone hotspot mean your team keeps working from home, a coffee shop, or anywhere else with power.

This is also the quiet argument for laptops over desktops. A laptop is a computer with a built-in UPS that runs for six hours. A desktop dies mid-sentence. If you are still weighing up the two platforms, our honest comparison of Microsoft 365 and Google Workspace covers the trade-offs in detail; for continuity purposes, either one moves your core systems off your premises and off the local grid.

Protecting the Servers You Still Have On Site

Some workloads genuinely still belong on premises. Those need proper protection, not just a plug strip with a battery in it.

• Pure sine wave, with communications: The server UPS needs clean output and a USB or network connection to the machines it protects. Without the data link, it is just a countdown timer.
• Automate the shutdown, then test it: Configure the shutdown software, then pull the mains plug during a quiet window and watch what actually happens. Shutdown scripts fail silently far more often than anyone expects.
• Condition the power coming back: Surge protection and automatic voltage regulation matter most at restoration, when the grid is at its dirtiest. Bring servers back up after the network gear, not with it.
• Be careful with generators: Cheap portable generators produce power rough enough to damage sensitive equipment. If a generator feeds your server room, put an online double-conversion UPS between them to clean the supply.
• Back up as if corruption is inevitable: Power events are one of the leading causes of on-premises data corruption. Follow the 3-2-1 rule, keep one copy off site or in the cloud, and test restores, not just backups.

And keep asking the harder question: does this server still need to exist? Every workload you move to the cloud is one less thing to keep alive when the lights go out.

The Priority Checklist for an SME Budget

If the budget is tight, spend it in this order. Each step delivers value on its own, and none of them requires the one after it.

• 1. UPS the network rack: R1,500 to R5,000. The cheapest, highest-impact move available. Everyone with a charged laptop stays online through the entire outage.
• 2. Add LTE failover: A failover-capable router and a data SIM. Covers the outages a UPS cannot, for less per month than a single hour of company-wide downtime.
• 3. Move email and files to the cloud: If you already pay for Microsoft 365 or Google Workspace, you may be most of the way there. Make sure files actually live in SharePoint, OneDrive or Drive, not on a PC under someone's desk.
• 4. Buy laptops on the next refresh: Same money you were going to spend anyway, and every desktop you retire is one less machine needing backup power.
• 5. Protect any remaining server properly: Pure sine wave UPS, automated shutdown, tested twice a year.
• 6. Add a lithium inverter for the workspace: Screens, lights and chargers for the team. For a small office this typically lands between R15,000 and R30,000.
• 7. Test the whole chain quarterly: Kill the mains during a quiet hour and watch what happens. An untested backup system is a hope, not a plan.

The Bottom Line

Power resilience is not about panic-buying a generator the week the stages return. It is boring, layered engineering: keep the network alive, shut servers down cleanly, keep a second path to the internet, and keep your core systems in the cloud so the building itself becomes optional. Most SMEs can put all of that in place for less than the cost of one lost trading day.

If you are not sure where your setup would fail first, it is worth finding out before the grid tells you. AMEA runs exactly this kind of resilience check for businesses: a short audit of where the single points of failure sit, and a prioritised plan that usually starts with a R2,000 UPS rather than a R200,000 generator.