← Back to blog

Everything Fails: Why Mission-Critical Systems Need Event Sourcing

The FAA ground stop showed what happens when legacy systems fail. Why mission-critical software should not run without event sourcing and a full history.

One week, multiple failures, and one lesson architects keep relearning

The Federal Aviation Administration’s recent nationwide ground stop, halting departures across the United States, caused confusion, disruption, and negative economic impact.

One day earlier, a widespread Microsoft 365 outage disrupted Outlook, Teams, Exchange Online, and other services for businesses and government operations around the world. One estimate from Anderson Economic Group put the resulting productivity losses at about $7 billion.

Later in the week, other US government agencies such as the National Security Administration, National Weather Service also reported outages. The outages extended beyond government agencies as well. Large banking institutions, airlines, hotels, and multinational technology companies also reported outages of some kind.

Even as unrelated events, it’s worth examining them together.

Werner Vogels has spent years reminding the industry that everything fails, all the time. November 2026 offered another demonstration of this at national scale. The question for anyone building software people depend on isn't whether a system will fail, because at some point it likely will. The question is whether that system can explain what happened, establish what remains true, and recover with confidence.

The Limitations of Traditional Data Storage for System Recovery and Auditability

Systems that only store current state may not be able to explain how they got to that state or establish which information can still be trusted. Many enterprise applications work this way. A record is created, then updated, then updated again, with each update overwriting the one before it. The database holds an answer to what things look like right now, but often very little about the business decisions and changes that produced that answer.

Under normal conditions, this limitation can remain invisible. Under failure, it becomes a serious problem. When an outage hits a state-based system, operators may have to reconstruct the past from fragments such as application logs, partial backups, the recollections of whoever was on call, or timestamps on support tickets. Recovery turns into forensics, and forensics is slow precisely when speed matters most.

No architecture prevents every failure. Hardware wears out, people make mistakes, and networks congest. What architecture determines is the quality of the evidence available afterward, including how much the system knows about its own history, how confidently it can be restored, and how clearly anyone can answer the questions regulators, executives, and the public will ask and sometimes demand next.

Those questions arrive quickly when millions of people are affected. Travelers, businesses, and lawmakers want to know what failed, when it started, what the system was doing in the minutes before, and how long recovery will take. Answering them requires a reliable record of a system's activity before, during, and after a failure, and most systems were never designed to keep that record.

The Risks of Legacy System Modernization in Mission-Critical Environments

Legacy systems carry a cost long before they fail. In September 2024, the Government Accountability Office found that more than 75% of the FAA's 138 air traffic control systems were unsustainable or potentially unsustainable (GAO-24-107001). Congress subsequently appropriated $12.5 billion toward the FAA's air traffic control modernization initiative, including $9 billion for systems modernization and $3.5 billion for facilities construction. The FAA estimates that Phase 2 of the systems modernization effort will require additional funding.

Modernization does not eliminate legacy risk immediately. It introduces a transition period in which existing systems must remain operational while replacements are built and deployed. During this period, organizations must maintain the reliability of infrastructure they are actively trying to replace, often while managing dependencies between old and new systems.

The FAA's history illustrates the consequences of this underlying fragility. In 2023, a failure in the NOTAM (Notice to Airmen) system, which alerts pilots to hazards along their routes, led to the first nationwide ground stop since September 11, 2001, delaying thousands of flights.

The cost of failures like these extends well beyond the hours a system is down. Rebooked passengers, repositioned crews, lost productivity, and disrupted operations compound the impact. Behind those disruptions are people who miss connections, appointments, family events, and time they cannot get back.

Mission-critical software is defined by the consequences its failure can have on the people, operations, and services that depend on it. In aviation, healthcare, finance, and public infrastructure, reliability isn't optional and the consequences on humans are real.

Why Legacy Infrastructure Puts Aviation (and more) at Risk

These challenges aren't limited to aviation infrastructure. Any organization operating across complex, interconnected systems needs to reconstruct what happened when something goes wrong. In aviation, that means understanding operational decisions and system changes. In financial services, it means establishing how a transaction was authorized. In food safety, it means tracing products through a supply chain.

In 2026, U.S. health officials investigated a large, multistate cyclosporiasis outbreak that sickened thousands and killed at least two people and found it was linked to iceberg lettuce. But even after the outbreak was declared over, questions remained about exactly how the contamination occurred.

Food safety and air traffic control have little in common on the surface, but both depend on the ability to reconstruct a precise sequence of events across interconnected organizations and systems. Investigators need to know where something originated, what happened along the way, and which decisions or changes contributed to the outcome.

Reliable, time-ordered business records can make that investigation faster and more precise. They do not replace physical traceability, supplier controls, or regulatory processes, but they provide the historical evidence needed to understand how products, transactions, and decisions moved through an organization. A recall can only be as fast and as targeted as the records supporting it.

Mission-critical systems should not run without event sourcing

Most software stores only its current state. When something fails, a backup can restore the data, but it cannot show how the system got there or which decisions led to the problem. Traditional observability looks at technical logs and not specific business logic that triggered said actions and history. This leads to a forensic exercise.

Event sourcing takes a different approach. Instead of overwriting the past, an event-sourced system records each meaningful business change as an event, in the order it happened. Current state is derived from that history, so the record of what happened is never lost and is articulated based on your business rules and logic

That difference changes what an organization can do on its worst day in four ways.

1. It can trace what happened. Teams start from an ordered record of business changes rather than piecing together a timeline from scattered technical records and individual memories. When the inputs behind each decision are captured alongside the events, investigators can see not only what changed but why, and pinpoint where a process went wrong.

2. It can rebuild state from history. If application state becomes corrupted, teams can replay events from a known-good point to rebuild it. This adds a recovery option beyond restoring the last database backup. It does not replace backups, replication, or tested recovery plans, and the event store itself must be protected, but it means recovery no longer depends entirely on the condition of a single backup.

3. It can support modernization without duplicating business history. Modernization usually means running old and new systems side by side, each holding its own version of the truth. With event sourcing, existing and new components can work from the same record of events, so teams can introduce and validate new functionality step by step instead of replacing everything at once. Integration still takes work, but teams spend less effort keeping separate copies of the same history in agreement.

4. It can provide evidence for business decisions. Regulators, auditors, and executives increasingly expect organizations to explain how important decisions were made. An event history shows what information was available at the time and whether the resulting action followed the intended rules. That matters even more as AI agents begin making operational decisions inside critical systems, because deciding whether an agent can be trusted to act on its own requires knowing what it saw and why it acted.

None of this means event sourcing prevents every failure. Hardware will still wear out, configurations will still be pushed by mistake, and networks will still go down. But event sourcing can do more than preserve a record of what happened. Combined with explicit business rules and reliable workflow orchestration, it can help prevent invalid operations, enforce process requirements, and contain failures before they cascade. And when something does go wrong, the ability to understand what happened, reconstruct the decisions that led there, and recover from a reliable history becomes essential.

For systems where failure affects people's safety, finances, or access to essential services, these capabilities should be a baseline requirement, not an afterthought.

Aviation already works this way. Flight data recorders do not prevent accidents, but they make it possible to understand them. Mission-critical software deserves the same commitment to preserving the history of its decisions.

How to Modernize Mission-Critical Systems Without Disrupting Operations

The hardest part of this argument is not the technology. Most mission-critical systems were built long before event sourcing became a common architectural pattern. The people responsible for them cannot simply shut down operations for a multi-year rewrite. Running old and new systems side by side, under schedule pressure and with complex dependencies, is where modernization programs become especially fragile.

The path forward has to start with the systems organizations already run, not with a blank page. And organizations do not need to rebuild everything to start capturing the business events that matter.

The Axoniq Platform gives enterprise teams the infrastructure and development patterns to build event-driven applications.

For organizations running legacy systems, the first step is seeing what those systems actually decide. Axoniq Clarity, now in private preview, records the business decisions a running application makes and keeps the evidence, without a rewrite or code changes from legacy systems.

Axoniq Clarity runs on customer-controlled infrastructure with read-only access, keeping all data within the organization’s network. It creates a living record of what running applications decide and why, giving teams a practical starting point for introducing event sourcing incrementally.

Adopting event sourcing does not require rebuilding existing systems. It starts with understanding the business events, decisions, and dependencies already inside them.

These outages will be studied, debated, and eventually addressed. But the next failure is already waiting somewhere in a system that preserves only its present state. For organizations running software people depend on, preserving the history of business decisions and actions is no longer simply an architectural preference. It is a matter of operational resilience and regulatory requirements.

Request access to Axoniq Clarity, or talk with our team about introducing event sourcing into your existing applications.

Frequently asked questions

What did the November 2026 FAA ground stop cost consumers?

More than 18,500 flights were canceled during the 2026 FAA ground stop. An Anderson Economic Group analysis estimated that passenger costs from cancellations and delays exceeded $1.9 billion. Separately, Bernstein estimated roughly $345 million in pretax losses for the four largest U.S. airlines alone, while Transportation Secretary Sean Duffy said the broader economic cost ran to billions of dollars.

Can event sourcing prevent outages like the FAA ground stop?

Event sourcing does not prevent hardware failures or configuration mistakes full stop. It can help orchestrate workflows based on business logic, native to the code. It also changes what happens afterward, giving operators an exact record of what the system did, the ability to rebuild state from a known-good point, and a safer way to run old and new systems side by side during modernization.

Why does event sourcing matter for legacy system modernization?

Modernization is riskiest when old and new systems must run in parallel and agree with each other. Event sourcing lets both consume the same sequence of events, so teams can build and verify new components incrementally without relying on fragile database synchronization or a full rewrite.

Which industries need event sourcing most?

Mission critical industries need event sourcing the most. Any industry where software failure directly affects people benefits most, including aviation and transportation, healthcare, financial services, government, logistics, and food supply chains. These sectors also face the strictest expectations for traceability and auditability.

See it in production

Otto, Nets, MoneyLion, DPG Media and other teams run event-sourced systems in production with Axoniq.

Read customer stories

Join the Thousands of Developers

Already Building with Axon in Open Source

Talk to us about LTS →

Join the Thousands of Developers

Already Building with Axon in Open Source

Talk to us about LTS →