The FAA Ground Stop Shows Why Network Transformation Must Be Engineered For Failure
A change at two locations reportedly destabilized communications across 17 of the US Federal Aviation Administration’s (FAA) 22 en-route control centers. This is more than an outage story. It’s a warning about what can happen when network modernization expands the potential blast radius faster than the organization’s ability to observe, contain, and recover from failure.
On November 13, the FAA halted departures across the United States after instability in its communications infrastructure affected systems supporting air traffic control. The nationwide ground stop lasted more than four hours. Most recently, US government and industry officials said the incident emerged during work to migrate connectivity from legacy copper and Time Division Multiplexing (TDM) services to fiber and IP-based telecommunications. They reported that unusual traffic associated with newly installed Juniper routers in El Dorado, Arkansas, and Amarillo, Texas contributed to instability across the broader network.
The investigation remains underway, and public accounts differ on the exact combination of equipment behavior, configuration, installation, network design, and operational decisions involved. The final root cause may change, but the architectural and operational questions raised by the incident remain.
The central lesson may not be that an old network failed or that a new router misbehaved. It is that network transformation itself must be engineered for failure.
The FAA Network Is More Familiar Than It Looks
The FAA network may appear unique, but it shares many of the same characteristics and risks found in large enterprise networks.
- It’s Hybrid By Design. Like many enterprises, the FAA operates modern IP infrastructure alongside legacy technologies, such as TDM, that still support critical operations. Manufacturers, utilities, transportation providers, hospitals, hotels, and building operators face the same challenge as they connect modern IT environments with older OT and purpose-built systems.
- Legacy Does Not Mean Unimportant. Older infrastructure often supports the most consequential business and mission functions. Modernization must maintain operations while old and new environments coexist, sometimes for years.
- It Is A Business Asset. The FAA network enables safety, airline operations, passenger experience, cargo movement, and economic activity. Its success must be measured against those outcomes, not only device and circuit availability.
- Network Failures Become Business Failures. Delta’s 2016 data-center outage reportedly cost about $150 million. Southwest’s 2022 disruption produced roughly $1.18 billion in pre-tax impact, followed by a $140 million Department of Transportation penalty. Those incidents affected individual airlines; an FAA disruption can expose the entire aviation ecosystem at once.
- It Is A Business-wide Fabric. Data centers, campuses, branches, clouds, edge locations, and operational environments are no longer independent networks. A device at a small or geographically remote site can participate in services with nationwide dependencies.
- Remote Does Not Mean Low Risk. Business criticality should be determined by dependency and potential blast radius, not location, facility size, or employee count. The reported impact across 17 control centers makes that distinction difficult to ignore.
- It Depends On An Ecosystem. FAA teams, carriers, contractors, equipment vendors, and systems integrators all contribute to the service. Tasks may be outsourced, but accountability for safety and continuity cannot be.
- Redundancy Does Not Guarantee Resilience. Backup circuits and devices can still share carriers, conduits, facilities, control planes, or operational dependencies. Two links do not necessarily provide two independent paths.
- It Must Modernize While Operating. The FAA cannot close the national airspace while replacing critical infrastructure, just as a bank, factory, hospital, retailer, or cloud service cannot stop operating during modernization.
- Complexity Is The Common Denominator. Like most large enterprises, the FAA is not dealing with one network, vendor, technology, or operating team. Resilience depends on how technologies, people, processes, suppliers, and mission systems work together.
The Failure Pattern Is Familiar, Too
The FAA incident is part of a broader pattern in modern enterprise networks: seemingly localized changes can propagate across shared infrastructure and interrupt entire industries. A routing-filter change at Rogers in 2022 disrupted connectivity for more than 12 million Canadians, including payment and emergency services. A defective CrowdStrike update in 2024 crashed approximately 8.5 million Windows systems, interrupting airlines, hospitals, banks, retailers, and government services worldwide. The 2025 Amazon Web Services disruption in US-EAST-1 generated more than 17 million outage reports and affected over 3,500 companies across more than 60 countries. Different technologies failed, but the business lesson was consistent: highly connected infrastructure can turn a narrow technical problem into a widespread operational event. Resilience therefore depends not only on preventing failure, but on limiting how far failure can travel.
Operational Failure And Cyberattack Share A Blast-Radius Problem
There is no public evidence that the FAA event was a cyberattack. However, accidental failures and malicious activity can exploit the same architectural weakness: excessive trust that allows unexpected behavior to propagate. A misconfigured device, compromised administrator, malicious insider, defective update, or vulnerable supplier can produce different technical events but the same business outcome.
Forrester’s Zero Trust research emphasizes continuous verification, granular policy enforcement, least privilege, and containment rather than assuming that anything inside the network is inherently trustworthy. Applied to critical infrastructure, this means authenticating administrators and systems, restricting management access, segmenting operational environments, monitoring policy and configuration changes, and isolating abnormal behavior before it crosses fault domains.
A resilient network contains both attacks and accidents. It should not matter whether abnormal behavior starts with a threat actor, a defective update, a configuration error, or a malfunctioning device; the architecture must prevent it from becoming a business-wide event.
Engineer Network Transformation For Failure
The FAA event offers practical lessons for any organization modernizing a critical network. These include:
- Plan For Failure, Not Just Change. Model dependencies, validate the change, define go/no-go thresholds, and rehearse rollback before touching production. Require privileged access management, strong administrator authentication, approved golden configurations, separation of duties, and an auditable record of who changed what, when, and why. Keep heightened monitoring and decision authority in place after the change.
- Schedule Around Business Risk. Choose maintenance windows based on operational exposure, weather, traffic, staffing, major events, and the organization’s ability to absorb a failure, not simply when engineers or contractors are available.
- Limit The Blast Radius. Use segmentation, fault domains, rate limiting, control-plane protections, policy enforcement, and architectural isolation. The 2025 Amazon Web Services outage showed how dependencies concentrated in one region can disrupt thousands of organizations worldwide. A local failure should degrade a service, not destabilize the wider business ecosystem.
- Treat Segmentation As Both A Security And Resilience Control. Segmentation should contain cyberattacks, malformed traffic, configuration errors, software defects, and unexpected device behavior. Define zones for systems with similar mission and security requirements, enforce policy at the boundaries, and prevent management traffic from propagating freely across operational domains.
- Observe Operational And Security Behavior Together. Correlate configuration changes, connection rates, routing behavior, device health, authentication, privileged activity, policy violations, and service dependencies with operational impact. Shared NOC and SOC context helps teams determine whether abnormal behavior is a failed change, equipment problem, cyberattack, or combination of events.
- Prove The Backup Is Independent. Validate physical-path diversity, carrier diversity, automatic failover, restoration procedures, and the absence of shared dependencies. Exercise the backup under realistic conditions rather than assuming it will work.
- Migrate Incrementally. Build the replacement alongside the existing environment and use canary deployments, limited pilots, and defined burn-in periods. The CrowdStrike outage showed how quickly a defective update can propagate when deployment scope expands faster than operational confidence.
- Validate Before Every Cutover. Confirm traffic has moved, policies have propagated, monitoring is active, failover behaves as intended, and rollback remains available before retiring the old connection or platform.
- Outsource Tasks, Not Accountability. Make the RACI explicit across internal teams, carriers, contractors, vendors, and systems integrators. Everyone should know who detects, decides, escalates, communicates, executes, and rolls back.
- Extend Zero Trust To Suppliers And Administrators. Contractors, vendors, and systems integrators should receive only the access needed for an approved task, during an approved window, on approved systems. Record privileged sessions, validate device and configuration state, and revoke access when the work ends. A trusted supplier connection should never create unrestricted trust across the network.
- Map Dependencies Continuously. Document how sites, devices, circuits, routes, applications, control centers, and operational processes depend on one another. Revisit the map as the architecture changes.
- Measure Resilience In Business Terms. Track recovery time, failover effectiveness, service continuity, operational impact, and blast radius alongside conventional availability and performance metrics.
These lessons lead to a simple conclusion: modern is not the same as resilient. Fiber, IP, automation, and newer networking equipment can improve performance, but technology alone cannot guarantee continuity. The last decade is filled with outages caused not by obsolete technology, but by software updates, configuration errors, routing changes, shared services, and automated changes propagating farther than expected. Modernization can introduce new dependencies while legacy and modern environments coexist.
Zero Trust principles such as continuous verification, least privilege, segmentation, and rapid containment can limit what a compromised, misconfigured, or malfunctioning component affects. The measure of success is not how much old technology has been replaced. It is whether the organization can absorb an unexpected change, contain the failure, recover predictably, and keep the operation running.
For business and technology leaders, the question is not “How modern is our network?” Instead, it’s “When something fails anywhere in our business-wide fabric, how much of the business fails with it?”
If you would like to discuss how to align networking strategy with business outcomes, resilience, and continuity objectives, submit an inquiry to inquiry@forrester.com and schedule a conversation with us.