Cloudflare outage exposes global internet fragility, says BMIT CTO

Tuesday’s Cloudflare outage, which temporarily brought down X, Spotify, ChatGPT and a number of local websites, has once again highlighted the vulnerabilities at the heart of the modern internet  –  a system increasingly dominated by a handful of major infrastructure providers.

Speaking to MaltaCEOs.mt, Gordon Bezzina, Chief Technology Officer at BMIT, said the incidents involving Cloudflare, Amazon Web Services (AWS) and Microsoft Azure in recent weeks “highlight a common theme of technical and operational vulnerabilities inherent in today’s large Internet service providers”.

Over the past decade, businesses have moved from on-premises systems to large cloud platforms, a shift that Mr Bezzina notes has created unprecedented interdependence. These cloud giants “have grown at an exponential rate, resulting in increased complexity of the underlying platforms and management systems.”

He explained that each of the recent outages stemmed from different technical faults  –  but all exposed structural weaknesses.

“The AWS outage was caused by a DNS malfunction originating in the DNS management automation at their North Virginia data centre. Azure’s outage, which occurred shortly after, resulted from a misconfiguration in its Azure Front Door content delivery network (CDN). Meanwhile, Cloudflare outage also involved network service errors,” he said. While Cloudflare is still investigating the root cause, its CTO has pointed to a “routine configuration change” as one of the possible triggers.

Mr Bezzina emphasised that these failures did not stem from underinvestment. “Evidence points to a highly interconnected, integrated, and interdependent system that can be brought down by a single failure or misconfiguration,” he said, noting that this “reflects on the fragility of complex, interconnected cloud systems.”

The consequences were global. “In all three instances, the impact was not localised; but cascaded and had a worldwide impact,” he stressed.

Although unrelated, the incidents “highlight an uncomfortable truth: A few service providers have become critical single points of failure for a massive proportion of the Internet.” This concentration, he added, means that “failures, even those originating from small errors or glitches, can cascade quickly and have widespread impact.”

Mr Bezzina said that while local hosting can mitigate some risks, “it does not fully eliminate reliance on large global providers like Cloudflare”. The lesson, he insisted, is clear: “As the old saying goes, a businesses should not to put all their eggs in one basket and resilience and redundancy must be priorities at the service-provider level.”

He explained that BMIT strongly advocates hybrid setups that combine locally hosted infrastructure  –  through its data centres or sovereign cloud platform  –  with global hyperscalers such as AWS or Azure. “For security, we recommend a layered approach using both native and third-party systems to maximise protection,” Mr Bezzina added.

As cloud dependence deepens, he said, businesses and service providers must accept that robustness, redundancy, and diversification are no longer optional  –  they are essential for keeping the internet running.