Legacy System Risks: We All Have an El Dorado Router

legacy system risk

Schedule a Portnox Cloud demo today.

Contents

On the afternoon of November 13, a contractor in a small building in southern Arkansas swapped old copper lines for fiber and installed a new router. Within hours, a nationwide ground stop had disrupted air travel across the country for nearly five hours. El Dorado was named for a legendary city of gold. It now has a modern claim to fame: being the place where one router ruined everybody’s Friday and put legacy system risk on the front page.
 

The router was only the spark

The new box set things off, but the network underneath it is why one remote site could ripple across the country. The FAA’s own explanation was that its legacy equipment became unstable when data began flooding the network.
 
How old is that legacy equipment? The Transportation Secretary said the routers were purchased in 2001 and were obsolete by 2006, and they are still in use in 2026. That is twenty years of running on hardware its own era had already written off.
 
It also isn’t the first time. In 2009, a misconfigured router swapped into the same network during scheduled maintenance took down the national flight plan filing system for about four hours. Same kind of change, same kind of network, and a lesson that was available seventeen years ago. Old systems make every change riskier, and that risk compounds quietly in the background.
 

Why avoiding change feels safe

Nobody decides to run a twenty-year-old network. It happens one short-term decision at a time.
 
It starts with a change that goes slightly sideways. Maybe an upgrade took longer than planned, or a patch broke something small and unrelated. The lesson the team absorbs is that touching things is how you get paged at 2 a.m. So the next upgrade gets scheduled for next quarter, when things are calmer.
 
Things are never calmer. The upgrade slips again, and now it’s two versions behind instead of one. Skipped versions pile up, the gap between where the system is and where it should be gets wider, and every missed release makes the eventual jump bigger and scarier. The documentation drifts. The person who remembers why it was configured that way takes a new job. Eventually the system becomes the thing nobody wants to touch, and “don’t touch it” becomes policy.
 
Security teams know this feeling well from on-prem network access control. When the platform is five versions behind, an upgrade stops being maintenance and becomes a project. On-prem software often can’t skip ahead, so five versions behind can mean five separate upgrades. A project like that needs an owner, a budget, a window, and a plan for what happens if it goes badly. Nobody volunteers for that, so it slips another quarter.
 
Avoiding change feels like the safe choice because the risk of acting is visible and the risk of waiting isn’t. A botched upgrade has a date, a ticket number, and someone to blame. A system slowly turning into a hazard has none of those, right up until it does.
 
El Dorado is that same slide at national scale. Routers purchased in 2001 were still in service in 2026, so when one routine swap went wrong, the older equipment around it couldn’t absorb the hit. Put off enough upgrades to keep things feeling safe, and a routine change ends up with nowhere safe to fail.
 

Break small, recover fast

The day before the ground stop, Microsoft 365 went down for much of the world, and the preliminary report traces it partly to a defect from a Windows update deployed back in July. Exchange Online couldn’t switch away from the affected gateway for scale reasons, so when that one front door jammed, email stopped for nearly everyone. Even well-run systems fail, so the realistic goal is a system that breaks small and recovers quickly:
  • Keep blast radius small, so a bad change hits one site or one group instead of everything.
  • Roll out in stages, with a pilot group before the whole fleet.
  • Know the rollback plan before the change starts, and test it.
  • Make changes small and frequent instead of rare and enormous.
  • Give every upgrade a named owner, because a project without one is a project that slips.

Pick your downtime

Every environment has an El Dorado router somewhere, a piece of aging infrastructure that everyone avoids because touching it feels dangerous. The question worth asking is whether the business can afford an extended outage, or would rather schedule a maintenance window.
 
A maintenance window comes with a start time, an end time, and a rollback plan, and everyone knows it’s coming. An extended outage skips all of that and shows up whenever it likes…usually a Friday afternoon.

Share

About the Author

Picture of Kate Asaff

Kate Asaff

Kate Asaff is a Technical Product Marketing Manager at Portnox with more than two decades of experience spanning networking, enterprise IT, and cybersecurity. Before moving into product marketing, she spent over 15 years at SolarWinds in technical support and program management, helping bridge the gap between engineering and the people who rely on technology every day. Today, she writes about network access control, zero trust, AI, identity security, and passwordless authentication for the practitioners who implement them.

About the Author

Picture of Kate Asaff

Kate Asaff

Kate Asaff is a Technical Product Marketing Manager at Portnox with more than two decades of experience spanning networking, enterprise IT, and cybersecurity. Before moving into product marketing, she spent over 15 years at SolarWinds in technical support and program management, helping bridge the gap between engineering and the people who rely on technology every day. Today, she writes about network access control, zero trust, AI, identity security, and passwordless authentication for the practitioners who implement them.

Related Reading

Network Access ControlSecurity Trends

You Acquired the Company. Did You Acquire Its Access Risk?

November 18, 2026
News

Nobody Tests the Parachute: Microsoft, the FAA, and the Failover That Made it Worse

November 17, 2026
Artificial IntelligenceSecurity Trends

DeepSeek and Ye Shall Find: What the DHS Ban Can’t See

October 29, 2026

New research: detection alone isn't enough for AI access.

X