When the web goes dark

In May 2025, a brief yet wide-reaching internet outage in South Africa – caused by a single point of failure in a core internet exchange – revealed just how exposed even mature digital ecosystems can be. Adil El Youssefi, CEO of Africa Data Centres, explains why redundancy has to be built into our digital infrastructure.

The interruption only lasted a few minutes, but with South Africa’s internet penetration rate standing at almost 79%, it significantly impacted thousands of businesses and millions of users. The ripple effect lasted well beyond the interruption, disrupting online classes, meetings, and even online streaming of the national budget speech.

This event served as a timely reminder that in digital economies, redundancy – including backup network connectivity, power supply, and data storage – is no longer a technical afterthought but a strategic imperative that ensures operations are seamless and services continue uninterrupted.

Are we overdependent on select pathways?

Internet exchange points (IXPs) form the backbone of regional internet performance, acting as central hubs where networks meet and exchange traffic efficiently. A key takeaway from this and other similar instances is that even the most robust and resilient systems are vulnerable to failure for various reasons. All of these can be mitigated if the appropriate backups are in place.

This wasn’t a failure of design or engineering. Instead, it served as a reminder of the risks associated with relying on a single IXP to carry the bulk of national traffic. This centralisation introduces risk: when that hub falters, the knock-on effects can compromise an entire nation’s connectivity.

The reality is that many South African internet service providers (ISPs), enterprises, and digital platforms are heavily reliant on a single IXP for their traffic flows. While this offers simplicity and perception of cost benefits, it also introduces risk, particularly when redundancy paths aren’t well-defined or contingency scenarios aren’t robustly tested. This is why multi-path connectivity and diversified peering strategies should be an integral part of every enterprise’s risk management playbook. 

Building redundancy into infrastructure DNA

This isn’t just about uptime; it’s about designing infrastructure that enables systems to perform acceptably even when part of the network is placed under stress. It’s also about allowing ISPs, cloud providers, fintech platforms, and enterprises to route traffic dynamically, with the intelligence to adapt to congestion or failure. 

At Africa Data Centres, a business of Cassava Technologies, redundancy is built into our infrastructure and our services. Multiple data centre sites enable seamless workload transfer across geographic locations. Electricity supply issues are addressed with solutions such as varied power feeds, on-site backup generation, and battery support. Network resilience is ensured through connections to various terrestrial and subsea cable systems, as well as cloud and carrier diversity. This provides clients with access to a variety of service providers and interconnection points. Africa Data Centres’ customers have access to all of South Africa’s IXPs, including Johannesburg Internet Exchange (JINX), Cape Town Internet Exchange (CINX), and NAPAfrica, ensuring that there are always fallback routes and alternative pathways in the event of an incident.

Prioritise resilience over centralisation

As South Africa advances on its digital transformation journey, the demands on our internet infrastructure will continue to grow. From AI-powered solutions and real-time analytics to remote healthcare, digital education, and content streaming, we are more dependent on stable, high-performance connectivity than ever before. 

While the most recent incident was resolved swiftly and transparently, and South Africa was soon back online, it should prompt every CIO, network architect, and digital strategist to reflect on their redundancy strategy. Are connections diverse enough? Is the business dependent on a single peering point? Can systems withstand a few minutes of disruption without missing a beat, or are they exposed to a single point of failure? 

The difference between disruption and resilience lies in foresight, powered by infrastructure that prioritises resilience over centralisation. The goal isn’t just avoiding the next outage; it’s about incorporating redundancy into every stage of the digital transformation journey.