The end of last week was an odd one, and quite possibly a warning of what’s to come. Microsoft was at the heart of this latest outage, but it’s far from an isolated incident. Google Cloud stumbled in 2025, taking a chunk of the internet with it. Amazon Web Services had its own wobble, and Azure’s Front Door followed just days later. Cloudflare went down that November and CrowdStrike’s faulty update in July 2024 grounded flights and froze hospitals around the world.
The pattern is hard to miss. The tech industry has followed a predictable cycle: first came the wave of outsourcing, then the push to centralise everything under a single banner, and finally the dangerous assumption that certain giants are simply “too big to fail”. The result? When one vendor sneezes, thousands of businesses catch a cold and grind to a complete halt.
I have no ill will towards Microsoft, I’ve been in this industry long enough to remember having to turn off the Windows Firewall on XP SP2 as it turned it on by default. IT wouldn’t be what it is today without them. But they, along with the rest of us, are victims of a “rockstar-driven” tech culture that prizes shiny trends over time-tested, best-practice engineering.
That’s why I’d argue the current industry design philosophy is fundamentally flawed. Companies should prioritise bringing every element of mission critical systems under their control. Self-hosting is a great place to start, and a solid backup and disaster recovery (DR) plan behind it is a must. As Galina Goduhina of ONLYOFFICE eloquently put it, “the cloud has made infrastructure someone else’s problem, but it hasn’t made downtime someone else’s consequence.”
So how did we fare last week? To be honest, we experienced no disruption at all, apart from receiving significantly fewer emails than usual. In fact, thanks to a robust backup strategy, we haven’t suffered a major outage in years. The only exception was a single ISP failure, which we fixed by installing a second line from a different provider. We aren’t a big company, we don’t have a huge budget, but we’ve achieved this resilience by being smart with our systems and how they are deployed.
You can’t stop your vendor having an outage, but you can decide how much of it becomes your problem. The worst time to learn you have no control is with a major project on the line and your vendor saying ‘we’re working on it.’
If your primary vendor went down for 24 hours tomorrow, what would your plan be?


