> ## Content Index
> Fetch the complete content index at: https://www.cybrsecmedia.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# What a Delayed Flight and a Weekend of Outages Taught Me
- URL: https://www.cybrsecmedia.com/what-a-delayed-flight-and-a-week-of-outages-taught-me/
- Published: 2026-11-16T13:06:50.000Z
- Updated: 2026-11-16T20:57:43.000Z
- Description: A delayed flight, a ground stop and a two-day Microsoft outage all failed the same way: dependencies nobody tested losing. Resilience as practiced, not just planned is an organization only defense.
- Author: George V. Hulme
- Tags: FAA ground stop, Microsoft 365 outage, Commvault Shift, CYBR.SEC.Community, Article

I got my own small taste of systemic fragility last week, sitting at the gate while my flight's delay stretched from minutes into hours. The aircraft was eventually proven fine. The crew was ready. However, the problem, as the gate agent later explained, was maintenance: they couldn't email status updates. Email was down, and with it went a routine that apparently had no fallback. A multimillion-dollar aircraft and a gate full of passengers sat idle because one workflow depended on one communication channel nobody had thought to test losing.

It was a trivial incident compared to [Microsoft's two-day disruption](https://www.cybrsecmedia.com/microsoft-365-outage-exposes-security-risks-beyond-cyberattacks/), the [FAA's nationwide ground stop](https://www.cybrsecmedia.com/faa-ground-stop-microsoft-outage-and-the-cybersecurity-fud-machine-a-k-a-calm-down/), or the retail outages that same week: but the nature of the failure was the same. A dependency, an untested assumption, and an operation that couldn't function through a failure.

These circumstances really shouldn’t happen in 2026.

In the days before my flight delay, I attended Commvault's SHIFT conference, where the people who work to make these dependencies resilient through failure shared their experiences and lessons learned. As I sat at the gate, getting increasingly frustrated, I had resiliency on my mind.

[Subscribe to the CYBR.SEC.Media newsletter](#/portal/signup/free)

[Alex Levy, CISO of BNSF Railway](https://www.cybrsecmedia.com/bnsf-ciso-alex-levy-on-agentic-ai-risk-baby-proof-your-environment/), one of the largest freight railroads in North America, always has resiliency on his mind. "An outage for us is an outage for the entire country,” he said a few days before the recent series of email and flight outages impacted the nation. “Most things in this country move by rail,” he continued. “So if you think about all of those \[fuel, oil, lumber, petrochemicals and grain\] suddenly becoming unavailable to everybody all at once, you can imagine the catastrophe that creates."

Substitute "email and identity" for "rail," and you have the Microsoft outage. Substitute "routine fiber upgrade," and you have the FAA's.

Levy's answer to mitigating such risks starts with identifying assets and services, ranking their criticality, and then understanding the blast radius of those assets. "How do we build environments that are within a certain blast radius," he asked, "so we know if there's an issue within a particular area, it's cordoned off?"

But knowing the blast radius on paper isn't enough. BNSF backs it with a robust testing discipline: tier-zero core applications are recovery-tested monthly, tier-one quarterly, everything else at least annually. The recovery runbook is a gate in their release pipeline: ship an update without refreshing the recovery runbook, and the release doesn't go out.

Asked whether he'd ever needed those plans for real, Levy responded: "Anybody who's sitting in a CIO seat or a CISO seat and said they've never had to do a restore would be lying to you. Hardware fails, developers make mistakes. No one's infallible."

Commvault’s Christopher Mierzwa, who runs resilience programs and tabletop exercises, described the exercise that would have saved my flight’s timeline. When he takes Teams and SharePoint away from a room of enterprise executives, the reaction is almost always the same: "'So wait, Teams is down.' “Yeah. Teams is gone. Email is gone. Identity's gone," he responds.

Then comes the question that separates resilience from paperwork checkboxes: "Do you use the out-of-band comms? Have you actually tried them?" Most haven't. One organization he works with keeps a few rotating pagers in service each month. That’s right: pagers, as the backup for when the first backup fails. That sounds quaint until email goes quiet and a maintenance crew has no other way to communicate an aircraft’s readiness.

The airlines, Microsoft, the retailers, the agencies that stumbled this month almost certainly all had recovery plans. What the outages revealed is how many of those plans required the failed system to be in place in order to execute: perhaps a runbook stored in the cloud that was down, a contact list living in the email system that wouldn't load, or a status update that could only travel through the very email channel that wasn't working. These are all scenarios Mierzwa stressed should be part of any tabletop exercise designed to test one’s ability to recover and be resilient.

On Tuesday, during his keynote, Commvault CEO Sanjay Mirchandani framed a mindset shift that is needed: "For decades, enterprises treated recovery as insurance. He meant a service one pays for, such as a backup that is tested semiannually at best, and everyone hopes they never have to use it. That made sense when attacks and mistakes moved at human speed," he said.

There’s no indication that the week’s outages following the keynote were caused by AI or any attack. They were all likely caused by poorly run routine IT processes and procedures. The fallout from the outages is the same nonetheless. This week showed, once again, that mistakes and misconfigurations that affect concentrated dependencies move fast, and their blast radius can be substantial.

What this week demanded was something different from recovery, because recovery assumes an organization gets to fall first: something goes down, the recovery plan kicks in, and things are restored and stood back up. Such sequencing is a luxury when the failure impacts shared dependencies, and half the tools in the recovery plan go away with it.

Resilience is broader and more important in today’s world of concentrated risk: the capacity to keep operating, even in a degraded way, while the failure is occurring and the organization can absorb the fault, route around it, and focus on restoring what broke. NIST's [formal definition](https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-160v2r1.pdf?ref=cybrsecmedia.com) describes cyber resiliency as the ability to anticipate, withstand, recover from, and adapt to adverse conditions.

During his SHIFT keynote, Mirchandani's shared his version of resilience that folds the four NIST keys to resiliency into being “tried and tested recovery." It’s not defined as backups on a shelf. It’s not untested plans in a binder. It’s a confidence, earned by practice, that the business keeps moving when a shared dependency becomes unusable. The next Microsoft outage, the next ground stop, the next email failure is not an “if” such incidents will happen again. It's a “when” such incidents happen again. Building a resilient organization is the best defense against the eventuality.

[![](https://storage.ghost.io/c/ab/67/ab676516-71e3-473d-8f73-9e0692f5aaee/content/images/2026/11/2027-Events-Banner-32.png)](https://www.cybrsecmedia.com/conference/)