DEV Community: Surmon.cloud

Microsoft 365 outage: how to avoid the single point of failure that paralyzes a business

Surmon.cloud — Fri, 13 Nov 2026 13:08:42 +0000

Originally published in French on surmon.cloud.

In short: To avoid a single point of failure, a business starts by mapping its dependencies, then diversifies its providers. France Num recommends favoring European solutions to keep better control over data and costs. A fallback plan with another provider protects the most critical tools.

> "Today, we must avoid as much as possible having a single point of failure that paralyzes the whole company because one server is down at a provider, however big it is."

The global Microsoft 365 outage of November 12, 2026 shows what a single point of failure costs: email, calendar and the tools that depend on them all stop at the same time. It is the right moment to move to a more resilient and more sovereign model.

Which services did the Microsoft 365 outage affect?

A major outage mainly hit Outlook and Exchange Online, the Microsoft 365 email service. According to Le Monde Informatique, incidents were reported in 74 countries and territories*1. According to Downdetector data quoted in the same article, more than 130,000 reports about Outlook were recorded in the United States at the peak of the outage*1. These figures are based on user reports: they give an order of magnitude, and the exact number of people affected remains unknown.

The duration is as striking as the scope. Nearly 11 hours after the disruptions began, Microsoft was still struggling to stabilize its infrastructure*1. The identified cause is technical: a process responsible for establishing secure connections on the servers was saturated*1. The vendor says it has no indication of a cyberattack.

Why is relying on a single IT provider risky?

A single provider becomes a single point of failure as soon as several of your tools rely on the same building block. On November 12, the email outage spread: features of Copilot, SharePoint Online, Planner and some administration tools were also affected*1. In Teams, calendars and meeting scheduling were hit the hardest.

The chain reaction even went beyond Microsoft. OpenAI and Anthropic reported malfunctions in their Microsoft connectors*1.

The internal mechanism of the outage illustrates the same logic. The unavailability of many servers shifted traffic to the equipment still running, which made the problems worse*1. When everything goes through the same place, the load concentrates instead of spreading out.

Can outages at large providers be measured?

Yes, and the providers themselves publish what you need to track them. Amazon provides a public dashboard where the history shows every AWS service interruption over the last 12 months*2, viewable without an AWS account. Microsoft offers a service health page in its admin center, with a tab that shows incidents resolved over the last 7 or 30 days*3.

For a business owner, these pages are useful material. You can record the frequency and duration of the incidents affecting the services you use. You move from an impression to a measured finding, specific to your activity.

How can a business reduce its dependence on a single provider?

The process starts with a map of your dependencies, then the diversification of your providers. This is the roadmap proposed by France Num, the French government's public service for the digital transformation of very small businesses and SMBs, in a guide published on November 9, 2026 and written by Alain Garnier, CEO of Jamespot*4. The guide recommends identifying your dependencies and diversifying your providers, favoring European solutions*4, to keep better control over your data and costs and to retain the ability to switch providers.

The same guide lists the criteria to check before choosing a tool:

The nationality of the vendor and its owners.
The location of the data it holds.
The jurisdiction it is subject to.

According to France Num, these criteria matter all the more when a business depends on a few providers for its most essential tools*4. Awareness is growing: the guide cites Kyndryl's Cloud Readiness 2025 study, according to which 71% of French executives say they are concerned*4 about this digital dependence.

For an SMB or a small firm, a few simple steps move things forward:

List the tools whose outage blocks your day: email, calendar, shared files, line-of-business software.
Identify those that rely on the same provider.
Plan a fallback for the most critical ones, with another provider, preferably European.
Check that you can export your data easily.

That last point is becoming more important. Starting January 12, 2027, the European Data Act (article 29) forbids cloud providers from charging switching fees: Regulation (EU) 2023/2854.

Why spread your digital tools now?

The timing is right. Spreading your tools, knowing your dependencies and choosing solutions hosted under a law you control gives you continuity and freedom of choice. A business that frees itself from the single point of failure keeps working when a giant goes down.

My take

Global outage of Microsoft's servers. A huge number of services are affected today, precisely.

It says a lot about how dependent businesses everywhere in the world are on a few very specific services from large American companies. It also raises the question of digital sovereignty, one of tomorrow's challenges.

This is where I position myself: I offer different solutions, with a more resilient approach better suited to small businesses. I actually target small businesses rather than large ones, given the capacity I currently have on my cloud. But yes, it is a real topic today, precisely.

My view is very simple. Today, we must avoid as much as possible having a single point of failure that paralyzes the whole company because one server is down at a provider, however big it is.

Security flaws keep appearing right now, and outages follow one another at AWS and elsewhere. So we are entitled to ask several questions. Will this become recurring? Is the risk measured well enough? Is it time to change the paradigm we have been waiting for?

> All the more so since the 2027 law will apply in Europe: it forbids charging for exporting cloud data when you switch tools. There is a whole issue at stake here.

Frequently asked questions

How long did the Microsoft 365 outage last?

Nearly 11 hours after the disruptions began, Microsoft was still struggling to stabilize its infrastructure.

How can I view the history of AWS outages?

Amazon provides a public dashboard that shows AWS service interruptions over the last 12 months. This dashboard can be viewed without an AWS account.

What criteria should I check before choosing a digital tool?

Check the nationality of the vendor and its owners. The location of the data it holds also matters. The jurisdiction it is subject to is the third criterion.

Are French executives worried about their digital dependence?

According to Kyndryl's Cloud Readiness 2025 study, 71% of French executives say they are concerned about this digital dependence.

Sources

*1 Panne mondiale de Microsoft 365 : Outlook perturbé pendant des heures,

*2 AWS Health Dashboard - AWS Health - docs.aws.amazon.com,

*3 Comment vérifier l'intégrité du service Microsoft 365,

*4 Comment réduire les dépendances numériques de votre entreprise ...,

Digital Sovereignty and AI: Taking Back Control of Your Data

Surmon.cloud — Thu, 12 Nov 2026 14:00:00 +0000

Originally published in French on surmon.cloud.

Your commutes, emails, files, and AI queries pass daily through foreign services. Digital sovereignty means reclaiming decision-making power over this data: where it lives, who can read it, and what is done with it. Good news: it is built through small steps, without changing everything at once.

Digital Sovereignty: What Are We Talking About?

For an individual or a company, digital sovereignty boils down to three questions. Where are my data stored? Which law applies to the entity holding them? Can I switch tools without losing everything? The more you can answer "at my place," "a law that protects me," and "yes," the more you are in control of your digital life.

With AI, the issue becomes more sensitive. Every document you entrust to an online assistant is sent to its editor's servers. Keeping control over AI therefore starts with keeping control over what you feed it.

Starting with Everyday Applications

The first step is the simplest: replace one application with another. For navigation while driving, Roole Map is a French alternative to Waze. It is free, ad-free, certified Origine France Garantie, and does not resell personal data*1. Its business model relies on the Roole automobile club's services, not on advertising.

The browser is the second step. Firefox blocks common trackers in its Standard mode, and Strict mode goes further*2. The change takes five minutes: Firefox imports your bookmarks and passwords.

Choosing a Host That Protects You

A server in Europe is not enough. What matters is the law that applies to the company managing it. In the United States, the CLOUD Act of 2018 allows authorities to obtain data held by a US provider, even when it is stored abroad*3.

Swiss companies offer another path. Proton, based in Geneva, offers encrypted messaging and storage. Since June 2024, its largest shareholder is a non-profit foundation*4. Infomaniak, also in Geneva, transferred the majority of its voting rights to a Swiss public utility foundation on May 13, 2026: the company can no longer be sold without the foundation's agreement*5. For your emails and files, this type of structure protects your choice in the long term.

Running Your Own Tools, Including AI

The next step is self-hosting: installing a service on your own machine rather than renting it. Docker makes this accessible. A service is installed with a single command, and your data remains on your disk.

To automate simple tasks, n8n can be installed this way on your own server, and its community edition is free*6. Its license allows use for your internal or personal needs, without reselling it as a service*7.

AI follows the same path. A tool like Ollama runs open AI models directly on a computer running macOS, Windows, or Linux*8. Your documents then stay at your place, and each question asked to the AI costs nothing more.

Can We Really Take Back Control?

Yes, step by step. Each step brings a concrete benefit. You depend less on online services and their price changes. You gain resilience, provided you keep a backup of your data. You also learn new skills, useful for your activity's cybersecurity: you protect better what you know.

For a freelancer or an SME, the stakes are real. Your client files, quotes, and know-how are your capital. Keeping them at your place is protecting them.

My Thoughts

It is a vast subject, and it is handled like eating an elephant: bite by bite. I started with simple steps, like Roole Map instead of Waze, Firefox instead of Chrome, then Swiss services like Infomaniak and Proton. Then, I trained myself in open source and Docker. Today, my automations, AI models, and services all run almost entirely at my place, and I produce as much as I want without paying per usage. Nothing revolutionary on paper. But for a freelancer or a growing company, owning your data changes the game.

Sources

*1 Roole Map: the #1 GPS app on the App Store, is it worth it? (iPhoneSoft, 03/14/2026),

*2 Firefox's ad blocker (Firefox.com),

*3 CLOUD Act Resources (US Department of Justice),

*4 Proton is transitioning towards a non-profit structure (Proton, 06/17/2024),

*5 European digital sovereignty: Infomaniak’s founder transfers control of his company to a foundation (Infomaniak press release, 05/20/2026),

*6 Host n8n (n8n documentation),

*7 n8n License, Sustainable Use License (GitHub),

*8 Ollama, running AI models locally,