Do you like this site? Remember to share it to all your friends on Facebook and Twitter!

Thursday, November 19, 2026

U.S. Nationwide Ground Stop on November 13, 2026: The Facts and My Theory

Last week, on November 13, 2026 (U.S. time), a major incident grounded all flights across the United States for over four hours.

The issue happened inside a private network run by the Federal Aviation Administration (FAA). This network carries critical air traffic control communications—including voice calls between controllers at different airports and radar data. During the incident, the network became extremely unstable, causing poor voice quality and unreliable radar feeds. To prevent mid-air accidents, the FAA ordered all planes across the country to stay on the ground. Flights only resumed more than four hours later, once the network was stable again.

Below, I will first summarize the facts currently available from media reports, followed by my personal technical theory on what went wrong.

Passengers viewing a JFK Airport departure board during the outage
By Sashimi-b - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/wiki/Special:Redirect/page/203209560

What Caused the Outage: Upgrading IP Network Routers

Everything in this section is based strictly on public news reports and official statements.

The FAA’s working network is called the FAA Telecommunications Infrastructure (FTI). The FAA outsources technical operations—including hardware, software, and physical circuit maintenance—to an outside contractor called L3Harris Technologies.

L3Harris has been transitioning the network off legacy Time Division Multiplexing (TDM) and old copper telephone lines onto modern IP routing and fiber-optic circuits. Both daily maintenance and this copper-to-fiber upgrade project for FTI are handled by L3Harris. (Note: This should not be confused with FENS—FAA Enterprise Network Services—a separate next-generation network contracted to Verizon, AT&T, and Lumen that is still being built and was not involved in this incident.)

U.S. Transportation Secretary Sean Duffy confirmed that the legacy local routers being replaced at roughly 2,000 FAA sites are Cisco 600 series models manufactured around 2001 (which were originally supposed to be retired back in 2006). Using 25-year-old, entry-level Cisco 600 routers on a mission-critical aviation network is shocking to anyone in the networking field. The main central ("core") equipment at FAA headquarters is also legacy Cisco hardware from the same era, though its exact models haven't been disclosed.

The new replacement routers at the local sites are made by Juniper Networks (owned by Hewlett Packard Enterprise, or HPE; exact models undisclosed).

Because thousands of local routers needed to be replaced, the upgrade was done gradually in batches. Prior to November 13, dozens of Juniper routers had already been installed and turned on smoothly without any service disruptions. The team scheduled the work during the daytime because the risk seemed very low; at worst, they expected a temporary hiccup at just the local site being worked on.

On November 13, 2026, L3Harris technicians were installing a new Juniper router at a radio communications site in El Dorado, Arkansas. Shortly after it booted up, the FAA began receiving reports of network trouble across the entire country.

At first, the FAA didn't know where the problem was coming from and worried it might be a cyberattack. Because the network was unreliable—risking broken controller audio or delayed radar data that could cause a crash—the FAA immediately halted all domestic flights while engineers investigated.

Engineers noticed that the central network was getting hit with over 1,000 connection attempts every four to five seconds. This constant flood of requests slowed the legacy Cisco equipment down more and more, causing data delivery to alternate between normal speed and extreme lag. The network wasn't completely dead, just dangerously unreliable.

They traced the flood of abnormal requests to two locations: the newly installed router in El Dorado, Arkansas, and a radar site in Amarillo, Texas (where a similar Juniper router had been installed about three weeks earlier). To be safe, L3Harris blocked traffic from both sites, and the FAA sealed up both Juniper routers and sent them to an L3Harris lab in Florida for forensic testing.

As soon as those two routers were isolated, the network issues vanished. As an added precaution, the FAA paused all future Juniper installations and disabled the "broadcast" feature on all 69 active Juniper routers already deployed across the country. Meanwhile, HPE pushed back against hardware blame, stating publicly that its Juniper router did not fail and "functioned as configured by the installer."

After double-checking that all systems were back to normal, the FAA finally lifted the nationwide flight ban.


My Theory: What Actually Went Wrong?

Since dozens of previous router upgrades using the same deployment process succeeded without a hitch, I don't think a random hardware defect caused this—which aligns with HPE’s statement that the router worked exactly as configured. Instead, here is my personal theory (just an educated guess for reference): a combination of broadcast traffic from the new configuration and the old central Cisco equipment getting pushed past its processing limit.

While news reports broadly called the flood "connection attempts," in networking terms, those were likely rapid-fire retry or broadcast messages—either from encrypted tunnels trying to rebuild themselves, or from routers repeatedly trying and failing to establish a basic "handshake" with headquarters.

Here is how that could happen:

  1. Broadcast traffic and frequent "check-ins": When the FAA disabled the "broadcast" function on all 69 Juniper routers afterward, it gave away a major clue. Suppose the installer’s template enabled a broadcast or dynamic routing feature over a virtual private tunnel (like GRE/IPsec) back to headquarters, alongside aggressive automated "health check" timers (like OSPF Hello or BFD timers) checking in way too frequently. 
  2. No built-in safety brakes on 2001 hardware: Every time a new router comes online and tries to sync up, it puts a noticeable spike of workload on the central router's processor—much heavier than routers that are already connected and running smoothly. Because the central Cisco hardware dates back to around 2001, it lacked modern built-in protections (such as control-plane rate-limiting) to stop one noisy local router from hogging all its processing power.
  3. The tipping point: Previous upgrades may have just gotten lucky and stayed right below the central router's breaking point. Bringing the El Dorado, Arkansas router online—combined with abnormal traffic coming from the Amarillo, Texas site—was likely the final straw that pushed the central processor over 100% capacity.   

Put all three together, and you get a snowball effect: once an overloaded 25-year-old router hits its limit, it starts delaying responses or dropping data packets at random. When the local routers didn't get a timely "I hear you" response back from headquarters, they kept sending repeat check-in and broadcast messages over and over—creating the exact 1,000-request flood described by the FAA. 


💡Because the FAA and FBI investigation is still ongoing—and engineers are currently analyzing the routers in a Florida lab—we will have to wait for their final official report to know the full truth. 


One more thing…

Cisco 600 Series Router (Cisco 677)
Photo Source (Thanks!):
https://www.facebook.com/share/p/1J6nQBQsxd/

In Taiwan, the old Cisco 600 routers were everywhere around the year 2000. Chunghwa Telecom (CHT), the largest telecom and internet service provider in Taiwan, installed Cisco 600 routers in homes for ADSL internet. Because they were small, round, and fit in the palm of your hand like a turtle shell, Taiwanese people nicknamed any ADSL modems "little turtles" (小烏龜) for quite a long time.


More videos on Show IP Protocols YouTube Channel.


No comments:

Post a Comment

Tip: you can also anonymously comment here.

Popular Posts