Load-Shedding and Your Server: A Gauteng IT Continuity Plan

Load-shedding has been intermittent since 2024, and that has let a lot of Gauteng businesses relax. The power still drops — just less predictably. Municipal outages, cable theft and substation failures in Tshwane, Johannesburg (City Power) and Ekurhuleni now take offices down as often as Eskom stages do, and they arrive without a schedule. If your office has a server or NAS, fibre, a card machine and VoIP or Teams, this is the plan to keep it working.

What Actually Breaks When the Power Drops

The drop and the return damage different things.

• The server or NAS does not shut down cleanly — Windows and the database are mid-write when the power vanishes. Usually it boots again; sometimes it does not.
• Database corruption — accounting, practice management and ERP systems keep open files. An unclean stop can leave a database missing the last hour of work, or one that will not open at all.
• RAID rebuilds — an array that loses power mid-write often marks a disk as failed on boot. The rebuild takes hours, and a second disk failing during it means restoring from backup.
• The fibre ONT and router go dark — the fibre tower down the road usually has battery backup. Your ONT and router do not, so the internet is up in the street and down in your office.
• Wi-Fi access points and VoIP phones — both run off the PoE switch. Switch off, Wi-Fi and phones off, even if every laptop still has battery. Teams on the laptop is fine until the Wi-Fi goes.
• Card machines — the countertop terminal needs power and a connection. Portable ones have a battery but still need Wi-Fi or a SIM.
• Printers — they survive, but the print queue on the server that just died does not.
• The UPS that beeps for eight minutes and dies — the battery in that old desktop UPS is good for a few minutes, and nobody has told the server to shut down before it gives out.

The Return Is Worse Than the Drop

Power does not come back gently. The voltage swings as the whole block restarts, and the spike arrives through the wall socket, the network cable and sometimes the phone line. A surge protector clamps the worst of it once, then it is spent, and most give no sign that the protection is gone.

A surge protector is not a UPS. It does nothing while the power is off and cannot shut a server down. You want both: the UPS to bridge the gap and manage the shutdown, and a surge stage — inside the UPS or on the distribution board — to take the hit on return.

The Four-Layer Plan

Layer 1: Know Your Block and Your Schedule

Gauteng is not one grid. Where your office sits decides who cuts your power and how you find out:

• Tshwane — the City of Tshwane runs its own schedule by block, and municipal faults are reported to Tshwane, not Eskom.
• Johannesburg — City Power runs its own blocks and outage reporting; substation failures and cable theft on its network are now as common as stages.
• Ekurhuleni — its own schedule and fault line, and parts of the East Rand sit on different blocks to the suburb next door.
• Eskom-direct — some industrial areas and townships take supply straight from Eskom and follow the Eskom schedule, not the municipal one.

Install EskomSePush, set it to your exact block (one suburb can straddle two) and turn notifications on. When a stage is announced you know the slots for the next few days; when it is a municipal fault, the app's area reports are often the first sign the outage is bigger than your building. We plan load-shedding per municipal block for every client site, because two branches of one business in Centurion and Kempton Park go off at different times.

Layer 2: Protect the Network Core First

If you buy one thing after reading this, buy a small UPS for the ONT, router and core switch. It is the cheapest, highest-value item in the whole plan. Those three devices draw a few tens of watts, so even a modest UPS keeps them up for a couple of hours, and the fibre tower is almost always still on. Internet, Wi-Fi (if the access points hang off that switch), card machine and phones all stay up.

Sizing is simple: add up the watts on each device's label, multiply by the hours you want to cover, and read the manufacturer's runtime table for the model you are considering. Do not guess — VA is not watts, and the table tells the truth. Put it where the ONT is, and label the outlets so nobody plugs a heater into it.

Layer 3: The Server — Shut It Down Cleanly, or Make It Matter Less

Option one is a proper server UPS with the shutdown software installed and its cable connected. The UPS tells Windows the power is gone, Windows waits a set number of minutes, then shuts down cleanly while there is still battery. That cable is the whole point — a UPS without the shutdown software is just a slower way to crash. We size UPS units for the router, switch, server and phones together, not just the desktops, because a server that stays up while the switch is off is a server nobody can reach.

Option two is to move the workloads that matter to the cloud so the server matters less. Emails, files and Teams already live in Microsoft 365 for most offices, and accounting systems increasingly have hosted versions. Once that is done, the on-site server is a file archive and a print server, and two hours off is an inconvenience rather than an outage.

Layer 4: People

• Laptops over desktops — a desktop dies the instant the power drops; a laptop runs for hours. Next time you replace machines, buy laptops and docking stations.
• Mobile hotspot fallback — if the fibre is down for the whole street, a phone with a data bundle and its hotspot on keeps two or three people working. Decide in advance whose phone.
• Teams on phones — install Teams on every staff phone, sign in and test it. Calls to the office number keep flowing and the customer never hears the power is out.

Generator, Inverter or Solar? Decision Logic for an Office

Once the core is protected and the server can shut itself down, the question is whether to keep the whole office running. That is a cost decision, and the maths goes like this:

1. List what must run: usually the network core, server, phones, card machine and a few laptop chargers. Not the aircon or the urn.
2. Add up the watts from the labels, or leave a plug-in power meter on the circuit for a day.
3. Multiply by the longest outage you plan to survive — a single slot for a scheduled stage, the working day for a municipal fault.
4. Compare with the manufacturer's tables: inverter and battery capacity in kilowatt-hours against your watts × hours; generator output in kilowatts against your peak load, plus fuel per hour.

An inverter and battery bank suits an office that needs the essentials for a few hours and can recharge from the grid between outages. Solar on top makes sense when outages are long enough that the bank would not otherwise recharge. A generator makes sense when the whole office, including heavy equipment, must run for a full day and someone will manage fuel, servicing and the changeover switch. Whatever you choose, the core and server UPS units stay: they bridge the changeover and still manage the shutdown if it fails.

The Office Manager's Load-Shedding Checklist

• EskomSePush installed, set to the office block, notifications on.
• Municipal fault line and account number on the wall.
• UPS on the ONT, router and core switch, tested this quarter, outlets labelled.
• Server UPS software installed, cable connected, shutdown timer set well below the runtime.
• Teams installed and signed in on every staff phone.
• Hotspot phone identified, data loaded.
• Card machine fallback agreed: portable terminal on battery, or a manual process.
• Surge protection checked after the last outage.
• Nightly cloud backup running, last restore test dated.
• IT support number on the office phone and on the wall.

Test the Plan Every Quarter

A plan that has never been tested is a hope. Once a quarter, switch off the mains at the distribution board for a few minutes and watch. Does the internet stay up? Do the phones ring? Does the server finish its shutdown before the UPS runs flat? Does the card machine still trade? Then bring the power back and confirm everything comes up on its own. Batteries lose capacity quietly — a UPS that gave you twenty minutes last year may give you five now, and the test is the only way to know before it matters.

How Backups Fit In

Everything above lowers the odds of losing the server. None of it lowers them to zero, and a server that takes several hard power cuts a year may not survive the year. That is why the backup is nightly, off-site and in the cloud — not on a drive plugged into the server that just lost power. A cloud backup from last night means the worst case after an unrecoverable crash is a day of work and a restore, not a business that cannot invoice. For recovery time, recovery point and what to restore first, read our disaster recovery page. If your website also lives in the office, proper hosting takes it out of the equation.

What a Managed IT Provider Does Here

You can do all of this yourself. What managed IT support adds is that it gets done and keeps getting done. From our side that means:

• Per-block planning for each client site, so the schedule and shutdown timers match where the office actually is.
• UPS sizing for the router, switch, server and phones, not just the desktops, with the shutdown software installed and tested rather than left in the box.
• Nightly cloud backup and a weekly on-site visit on the Fully Managed plan, which is when UPS batteries and the things that only show up in person get checked.
• Helpdesk 07:00 to 18:00 on weekdays, emergencies 24/7, because the outage that hurts is the one at 05:30 on a Monday.

Managed IT plans start from R750 per user per month; Fully Managed at R950 includes the weekly on-site visit and cloud backup. We work from 257 Jean Avenue in Centurion and support offices across Gauteng — see the areas we cover. If the server is already down, our IT emergency support guide covers the next ten minutes.