What size UPS does my office need for load-shedding?

For a typical 35-person office, a 1.5kVA online UPS dedicated to the comms rack (router, switch, access points) is the right starting point. It buys 30–60 minutes of network uptime. Add smaller units for reception and critical workstations. Keeping the entire office running on UPS alone is generally not cost-effective.

Is LTE backup internet worth it for a small business?

Yes, for almost any SMB where productivity depends on connectivity. A dual-WAN router with an LTE failover line keeps your office online during fibre cuts, ISP outages and most load-shedding scenarios. One prevented morning of downtime usually pays for the annual cost.

Do I need a generator for my office?

For most offices under 50 staff that can shift to remote work, no. UPS for the network, LTE failover for internet, and a documented work-from-home policy cover most outages at a fraction of the cost. Generators become worthwhile when customers physically visit, or once you cross roughly 50 on-site staff.

Why do my backups keep failing during load-shedding?

Overnight backup jobs are sized to run during a quiet window with the office powered up. Heavy stages cut that window short, leaving jobs incomplete. The fix is a cloud backup target that does not depend on office power, and monitoring completion daily rather than weekly.

Can my team work from home during stage 6 load-shedding?

Only if you have set them up to. Cloud-based email and file storage, offline-available critical files, a tested VPN where needed, and MFA on every account all need to be in place before the outage hits.

Load-Shedding Business Continuity for Johannesburg SMBs: An IT Operations Guide

It is 10:14 on a Tuesday in Sandton. Stage 4 hits without much warning. Inside a 35-person office, here is the order things fall over in: the lights blink, the airconditioning thumps off, three monitors go black because they were plugged into a wall socket and not into the UPS that was supposed to be feeding them. The receptionist's VoIP handset reboots and a call from a client drops. A designer's workstation, which is on the UPS, stays up — but the network switch in the comms cupboard was on a different circuit and was not. So the workstation is alive but offline. The fibre router is also on that dead circuit. Within forty-five seconds, twenty-eight people are sitting in front of working laptops with no internet and no shared drive. By the time power comes back two and a half hours later, you have lost the morning.

This is not a worst case. It is what an unplanned office looks like. The good news is that almost everything that breaks in that sequence is fixable for less money than most owners assume, provided you sequence the fixes correctly. This guide walks the IT operations view of load-shedding continuity for an SMB of roughly 10 to 100 staff in Johannesburg. It is the unglamorous middle: keep the business functional through a stage 4 to 6 day, without spending like a bank.

Start by mapping what actually has to stay alive

The first mistake most SMBs make is buying a generator before they have mapped their critical load. The right order is the opposite. Sit down with whoever runs operations and list every device, in priority order, that must remain functional for the business to keep earning during a two and a half hour outage.

For most JHB SMBs that list looks like:

Fibre router (ONT) and the primary network switch
The wifi access points serving working areas
A small number of mission-critical workstations (reception, ops desk, finance during month-end)
The VoIP handset or softphone gateway, if you still take calls on a desk system
Any on-premise server or NAS that is genuinely active during the day
The alarm and access control panel, if it does not have its own backup already

Things that almost never need to stay live: every staff laptop (they have their own batteries), printers, the office coffee machine, every monitor, the meeting-room television. Designing for only what matters drops your continuity cost by roughly half.

UPS sizing: what most offices get wrong

A UPS is a battery that buys you minutes, not hours. For an SMB the sensible job for the UPS is two things: keep the network alive while a generator starts up or while staff failover to LTE, and provide enough runtime for a clean shutdown of any on-premise server. That is it. Trying to run an entire office off a UPS through stage 4 is how you end up with a R180,000 lithium wall and still no internet.

For a typical 35-person office, a 1.5kVA online UPS feeding the comms rack (router, switch, two access points, VoIP gateway) will give you roughly 30 to 60 minutes of network uptime depending on load. Pair it with a smaller 650VA unit at reception and one per critical workstation. Total spend lands somewhere between R15,000 and R30,000 and removes the most common failure mode — the network dying before anything else.

Two practical notes. First, the UPS feeding your comms rack must be on a wall socket that is itself protected — a single dirty surge from the grid coming back online has killed more UPS units than load-shedding ever has. Second, lithium-iron-phosphate UPS units are increasingly worth the premium over lead-acid: longer life, no scheduled battery replacement every 18 months, and they survive being deep-cycled during heavy load-shedding stages.

Internet failover: fibre primary, LTE backup

Single-link fibre is the single biggest continuity risk in most JHB offices, and not because of load-shedding. Most fibre outages in 2026 are upstream — fibre cuts on the curb, ISP routing issues, a switch in the local cabinet dying. Whatever the cause, the answer is the same: a second internet path on a different carrier and a different medium.

The standard SMB setup is a fibre line as primary and an LTE or 5G router as backup. A dual-WAN business router automatically fails traffic over to LTE when fibre drops. Done correctly, your staff stay on the same wifi network, their Teams calls keep running, and most will not notice the swap. The maths is easy: if a single morning of forty staff sitting offline costs you more than the annualised LTE bill, the failover pays for itself in one incident a year.

Cloud versus on-prem: load-shedding picks a winner

This decision used to be a debate. Load-shedding settled it. Every workload you can sensibly move into Microsoft 365, Google Workspace or a managed cloud service should move. The reason is straightforward: when the office is dark, a Microsoft Teams meeting from your kitchen still works. Your on-premise file server does not.

The corollary is that your office is no longer the centre of gravity for your IT. It is just one of several places staff work from. If a stage 6 day means half your staff work from home, your continuity question is not "how do I keep the office open" but "is my team set up to work from anywhere when the office is not viable."

Microsoft 365 offline behaviour: the surprise gotchas

First, OneDrive and SharePoint files are not always available offline by default. Out of the box, files are stored as cloud-only placeholders. When the internet drops, those files are unreadable even though the document is right there in File Explorer. The fix is to mark critical folders for "Always keep on this device" in the OneDrive settings — proactively, for finance, ops and reception machines.

Second, Outlook in cached-exchange mode handles offline gracefully. Outlook on the web does not — refresh the page during an outage and you get the spinner of death. Make sure the desktop Outlook is the default on critical machines.

Generator economics for a 35-person office

A generator is the last resort, not the first. For most JHB SMB offices the maths only justifies a generator if you genuinely cannot move work to home during outages — typically because of customer-facing operations, physical inventory, or a contact centre. For most 10-50 person offices the more sensible architecture is: UPS for the comms rack, LTE failover for internet, OneDrive offline for critical files, and a work-from-home policy that kicks in automatically at stage 4 or higher. Total continuity spend stays under R50,000 and the business keeps earning.

Backup windows that survive grid instability

The forgotten part of continuity is whether your backups are still running cleanly. Heavy load-shedding stages disrupt overnight backup windows: the backup job starts, the office loses power before it finishes, and the next night the job re-runs from scratch, never completing. Over six weeks of stage 4-6 you can quietly drift into a state where your last good backup is from a month ago and nobody noticed.

Two fixes. First, move backups off-site and into a cloud target — Azure Backup, Wasabi, Backblaze B2, or your managed IT provider's offering. Cloud targets are not affected by your office power. Second, monitor backup completion as a daily check, not a weekly one. If your backup did not run last night, you want to know by 08:00 today, not when you need to restore.

A 7-step load-shedding audit you can run yourself

Block an hour, walk the office with a notebook, and answer each of these honestly.

Critical load map. Have you written down which devices must stay alive during an outage? If not, do that first.
UPS reality check. During the next outage, time how long the comms rack actually stays up. Compare to what the vendor promised.
Failover test. Unplug the fibre. Does LTE failover actually work? Do staff stay on the same wifi? If you have not tested this in 90 days, you do not know.
OneDrive offline status. On three critical machines, check whether the relevant folders are set to "Always keep on this device."
Backup completion. Pull the last 30 days of backup job logs. How many succeeded? If the answer is not "all of them," you have a problem.
Phone routing. When the office is dark, where do client calls go? If the answer is "voicemail," that is a revenue leak.
Work-from-home readiness. Can every staff member who can work remotely actually work remotely today, with VPN, MFA, and access to what they need?

If you can answer all seven cleanly, you are in better shape than most. If three or more are uncertain, you have meaningful exposure.

When DIY ends and managed IT begins

A small office with a switched-on operations manager can run most of this themselves. What is hard to do as a side-of-desk job is the ongoing discipline. Backups quietly failing in week three of stage 6. The UPS battery aging out and nobody noticing until it fails. The dual-WAN router quietly running on LTE for six weeks because the fibre cut was never reported, and your data bill is about to triple.

Continuity is not a project, it is an operating discipline. For a 10-50 person office, the crossover where handing it to a provider beats chasing it internally usually happens in month two or three of doing it themselves.

If you want a no-pressure conversation about where your current setup sits, contact CyberPeak. We will walk your office, run this audit with you, and give you a written priority list — including the items you can do yourself.