The FAA now says what stopped every departure in the United States on Friday, November 13: a router. A contractor swapped one in at a remote site in El Dorado, Arkansas, around midday, and by 3:54 p.m. Eastern it had flooded the network that links the country’s air traffic control centers.
I wrote about the ground stop on November 15, when the cause was still a guess. Here’s what has come out since, and the one sentence in it that I think matters most.
What the FAA says happened
The contractor was L3Harris, which is replacing the FAA’s old copper lines with fiber. The new router was a Juniper Networks device. Transportation Secretary Sean Duffy said it started hammering the other routers on the network. “We had 1,000 connection attempts every four to five seconds,” he said.
A second router, in Amarillo, Texas, had been replaced about three weeks earlier and did the same thing, according to Duffy. Officials think a feature that broadcast data more widely than it needed to caused the overload, and the FAA has switched that feature off on all 69 routers of that type. (Its chief technology officer said roughly 60 a day earlier. Yahoo News reported both figures and nobody has reconciled them.)
The trouble first showed up as static at the Indianapolis center. It spread to 17 of the FAA’s 22 en route centers, the ones that handle planes between airports. Nick Daniels, president of the controllers’ union, put it plainly:
“We know we didn’t have communications with airplanes.”
Tracing the problem took about 45 minutes. FAA Administrator Bryan Bedford said one reason is that maintenance work is logged on paper.
Nobody has said whether the router was faulty or set up wrong. HPE, which owns Juniper, says its devices “did not fail.” Both routers are now in an L3Harris lab in Florida, and the FBI is involved.
What it cost
The counts I gave on November 15 were early, and they were low. FlightAware logged 9,201 cancellations on Friday and 6,697 on Saturday. The FAA’s own figure for Friday is about 12,462, of which 9,082 were major airlines. By late Sunday afternoon, two days after a stop that lasted four hours and 21 minutes, close to 2,000 more flights had been canceled.
The sentence that matters
Every facility relies on one network, Bedford said, and
“there’s no backup to the backup.”
He added that the network “will continue to break until we replace it.” Duffy said more disruptions are possible until the last copper line is gone, which is scheduled for the end of 2028. L3Harris says the job is more than 70 percent done.
So there are about two more years of swapping parts on a live system that has no spare. And this has happened before. In 2009, a technician for Harris Corp., a predecessor of L3Harris, misconfigured a router during a fiber move and cut service to 129 air traffic facilities, according to a Transportation Department inspector general review that Yahoo News cited. The same kind of job produced the same kind of failure 17 years apart.
The upgrade has to happen, and I don’t blame anyone for doing it. What bothers me is the design: one box in Arkansas could reach nearly every center in the country, and we found that out on a Friday afternoon with planes in the air.
What you can do with this
If your flight was canceled, or delayed by three hours or more, and you chose not to travel, federal rules entitle you to a full refund. Airlines generally don’t owe you a hotel or meals for a stop the FAA ordered. The Transportation Department’s own site has the details.
Until 2028, I’d plan any trip that matters with a day of slack. And I’d keep what you need on a bad travel day (your bookings, a map, the phone numbers that matter) somewhere that works with no connection. That’s the idea behind the kit we sell and the free guide to building one.
The systems we lean on fail from the inside far more often than anyone attacks them. Our running list keeps getting longer.


