Resilience

Keeping operations running through
load-shedding: a systems view

Generators keep the lights on. They do not keep the business running if the systems it depends on are sitting in a server room with no power.

· 5 min read · GEMIS

The server in the cupboard

When the power goes, three things happen to a business that runs its systems on-site: the server goes down, the internet connection goes down, and the staff who could work from their phones cannot, because the system they need is the server that is down.

The usual answer is a bigger UPS or a generator for the server room. That keeps the server alive, but it does not restore the fibre, the router or the branch forty kilometres away that was also in stage four.

Move the system, not the generator

A system running on Microsoft Azure does not notice that your office has lost power. It is in a data centre with redundant supply, and it is reachable from any phone with a signal. The question becomes how your people reach it — and a mobile data connection is a great deal cheaper to keep alive than a server room.

For most of the businesses we work with, moving the core system to the cloud turned load-shedding from a stop-work event into an inconvenience. The office goes dark; the quotes, the orders and the schedule carry on from a laptop on a battery and a phone on LTE.

Design for the field as well

The office is only half the problem. A driver, an inspector or a technician in the field is often somewhere with no signal regardless of the grid. A mobile app that has to be online to work stops the moment they walk into a basement.

Offline-first design — capture on the device, sync when there is signal — covers both load-shedding and the ordinary realities of South African connectivity. It is a business decision made at the scoping stage, not a feature bolted on later, and it is the reason several of the field systems we built keep working on days the grid does not.

A short resilience checklist

  1. Is the system your staff need to do their jobs reachable from a phone, from anywhere?
  2. If the office loses power and internet for four hours, what can and cannot happen?
  3. Can field staff capture work without a connection, and does it sync itself afterwards?
  4. Are backups off-site, automatic and tested — not on a drive in the same cupboard?
  5. Who knows how to keep trading when the usual tools are unavailable, and have they practised it?

In practice

The case study behind this article

Sonke’s field app was designed for no signal at all: it captures offline and syncs when it can.

Non-profit — gender-based violence and HIV prevention

Sonke Gender Justice

Monitoring and evaluation across Africa

GEMIS’s role
Designer and builder of the monitoring and evaluation system
Status
Delivered 2018
  • Field operations
  • Integration

Keep reading

More from GEMIS

POPIA and your business systems: a practical checklist

Compliance is not a policy document. It is where your customer data lives, who can see it, and what happens when someone asks for it back.

  • Compliance
  • Data protection
  • Access control
  • Microsoft Azure

Tell us what your business needs to do better.

Discuss a project