Inspired by the 2021 Meta outage – The day communication itself became the incident
It started like any other morning at the city’s emergency coordination center. Calls came in, dispatchers relayed instructions, and fire, ambulance, and police units fanned out across the grid.
Then, almost imperceptibly, something shifted.
A few responders began reporting silence on the line — no acknowledgments, no directions. Others tried switching channels, thinking it was interference. Soon, chatter across all frequencies faded. The usual background noise of overlapping calls went still.
Within minutes, it became clear: no one could reach dispatch.
Crews were out in the field, responding to emergencies — but the very network that connected them had gone dark. They couldn’t coordinate, couldn’t confirm which calls were real or duplicate, couldn’t even check if help was already on the way.
Back at the center, the dispatchers faced the same silence. Their screens showed units active across the city, but they couldn’t reach any of them. The system meant to keep everyone connected had cut them off from the people depending on it most.
It wasn’t malice or sabotage — just a change meant to improve performance, rolled out to the same system that all communication depended on. There was no backup frequency ready, no separate control link to recover from inside the outage.
For a few long hours, everyone waited — professionals trained for chaos, suddenly unable to coordinate through it.
The incident beneath the story
In 2021, a routine network change at Meta disconnected the company from itself.
An update to the routing system that directed traffic across the internet also severed internal access, taking down communication tools and data centers alike.
The teams who could fix it were locked out — the digital version of responders unable to reach dispatch.
Recovery required physical access, out-of-band workarounds, and a reminder: when everything depends on the same system, resilience collapses quietly, all at once.
Lesson
Reliability isn’t just about keeping systems online.
It’s about preserving the ability to coordinate when they aren’t.
No matter how advanced our automation or how elegant our architecture, we still need a way to reach each other when the lights go out.
Reference
- Meta’s October 2021 outage postmortem (engineering.fb.com)
- Independent analyses: Cloudflare, ThousandEyes, and Kentik network reports
Next episode
Who knows ? But as most of us who work in technology know, it’s not a matter of if but when the next outage occurs. No system is perfect. We try our best and we prepare.