Comment | Resilience requires independence

November 16, 2026

The global Microsoft 365 outage exposed the risks of relying on a single provider. Now is a good time for organisations to take a clear-eyed look at how well prepared they are. Martin Hager, founder and CEO of Retarus, argues for better visibility into day-to-day operations, a robust IT architecture that keeps critical processes separate, and a fallback outside the primary provider’s platform. Customer choices help create monopolies. By choosing a broader range of suppliers, organisations can support competition and keep their own services running.

Last week, several Microsoft 365 applications, including Outlook and Teams, went down. Microsoft’s Azure cloud platform had already suffered an outage the previous month. Disruptions like these can affect any provider and will happen again. What matters is how far they spread. The real problem with the Microsoft 365 outage was that it affected all services, customers and regions at the same time. When consumer, enterprise and government services all run on a single shared global platform, it is hard to speak of digital sovereignty. These recurring outages show just how important resilience and technological independence have become.

The impact is particularly acute when business-critical email is disrupted. Even after Microsoft resumed processing messages, delivery was still delayed in some locations.

Organisations that detect disruption early can respond more quickly. To do that, they need a clear view of how their services are running and access to relevant operational data in real time, ideally through their own monitoring systems.

Good architecture keeps failures from spreading

Many organisations rely on their provider to distribute services across several data centres. That offers protection if individual servers fail. In Microsoft’s case, however, an unusually large number went down at the same time. The extra load was shifted to the remaining servers, which were then overwhelmed. What matters is not simply having several data centres, but whether they can operate independently. Data centres that run in parallel on separate networks, each capable of handling the full load on its own, offer much greater resilience. This is known as an active-active configuration.

Keeping individual processes separate is just as important. Inbound and outbound communications and software updates should operate as independently of one another as possible. Asynchronous processing can further reduce dependencies between systems and help contain disruption.

A provider retains the most control when it owns and operates its entire technology stack, even though doing so is costly and demanding. For years, that level of caution was seen as excessive, or even paranoid. Today, it is proving to be a strategic advantage.

A plan B outside the primary platform

Redundancy does not guarantee independence. Organisations also need a fallback system that operates entirely separately from their primary provider. In normal operation, it keeps address books and email histories up to date in the background. If the main system goes down, staff can open their fallback mailbox in a web browser, including on a smartphone. They can still be reached at their usual email addresses.

A backup with another major provider subject to the same jurisdiction simply trades one dependency for another. Organisations can retain greater control by using certified data centres in the EU and ensuring that their data is not subject to access by authorities outside Europe. In an outage, digital sovereignty becomes a practical matter.

Preparedness is also a regulatory requirement

The NIS2 Directive requires many organisations to safeguard their communications against disruption. The Digital Operational Resilience Act (DORA) sets similar requirements for the financial sector.

There will be another outage. The organisations best placed to weather it will be those that spot it early, have systems designed to contain the disruption and can keep communicating without waiting for their provider to fix the problem.

Related Articles

Engineered disorder could make protective materials more resilient

Researchers at MIT and Northeastern University have developed a metamaterial with an irregular network structure that withstands impacts better than conventional ordered lattices. The findings could inform future helmet and body armour designs, although testing has so...

IAA MOBILITY 2027: Early bookings signal strong demand in Munich

Around ten months before the event opens, exhibitor registrations are well ahead of the comparable point in the previous cycle. The 2027 programme will put AI, autonomous driving, battery technology and industrial transformation in focus. By 12 November 2026, 166...

Netherlands: Operation Endgame targets three malware networks

International investigators have taken down infrastructure linked to Gootloader, CastleRAT and StealC. The Dutch police report action against 14 Gootloader servers, six CastleRAT servers and 80 StealC servers. Working with international partners, they also seized...

Share This