
Outage resilience was tested three times in five days, as email, air travel and online banking all failed in the United States between Thursday 12 November and Monday 16 November. Nobody hacked anything. Each organisation was brought down by its own change, its own hardware or its own ageing kit.
Boards should find that more alarming than a breach, because the same conditions sit inside most large organisations right now.
Five Days, Three Failures
Shortly before 1pm US Eastern Time (6pm GMT) on Thursday 12 November, Microsoft 365 began failing across 74 countries, and it stayed down for more than 24 hours. Hospitals, universities and rail operators lost Outlook and Teams in the middle of a working day. Microsoft’s preliminary post-incident report traced the fault to a defect introduced in July during a network capacity upgrade. The update passed load testing and ran without trouble for four months, until a later network change brought the defect to the surface.
On Friday evening, 13 November, US Eastern Time, the Federal Aviation Administration, the US agency that runs air traffic control, grounded every departing flight in the country for more than four hours. It was the longest nationwide flight suspension since September 2001.
Investigators have traced the overload to a new router installed by L3Harris, the FAA’s network contractor, at a remote site in El Dorado, Arkansas, which flooded the network with connection attempts. The root cause is still under investigation, and the FAA has paused router installations while it finds out. Sean Duffy, the US Transportation Secretary, said the older routers on the network were bought in 2001 and obsolete by 2006.
From around 9:30am US Eastern Time (2:30pm GMT) on Monday 16 November, customers at Chase, Bank of America, Wells Fargo, US Bank and Truist reported problems with online banking and payments. JPMorgan Chase, the largest US bank by assets, blamed a software bug in an on-premise hardware system. The bank said other banks slowed only where they were party to a transaction with it. One bank’s bug still slowed payments for customers at other banks.
Availability Is Security
None of these was a cyberattack, and every one of them was a security failure. Confidentiality, integrity and availability have sat together in the security triad for decades, yet most boards fund the first two with far more urgency than the third. If your customers could not pay, fly or send an email, you had an availability incident, and availability sits on the security agenda whatever the root cause.
Look at the causes again. Two began with an upgrade. The third came from inside the bank’s own data centre. Each organisation broke itself while doing the routine work of keeping systems running.
Your Old Routers
Most organisations have a version of the FAA’s routers. Everyone knows the system is past its life, and it survives because replacing it never wins the budget round, the same quiet-year logic that puts security budgets first in line for cuts. Then somebody modernises the system next to it, and the old kit gives way under a load nobody tested for.
The Microsoft case adds a harder lesson. A change can pass every test, run cleanly for months and still cause an outage. Change management that stops watching once a release goes live will miss exactly that kind of failure.
Attackers Read the News Too
Outage weeks are good weeks for attackers. Within hours of the bank problems, social media filled with claims that US banks were under attack, and Chase was the only major bank to get ahead of that speculation publicly. Confusion like that is ideal cover for emails promising to restore an account or re-verify a card, aimed at staff and customers who have just spent a day locked out.
Outage Resilience: What to Do While the Board Is Still Listening
Start by finding your old equipment: list every end-of-life system still carrying critical work and put a replacement date against each one. Treat every upgrade as a security event, test how new work behaves alongside the oldest kit it touches, and keep a rollback ready long after go-live. Hold contractors to the same change discipline as your own staff.
Then test the response itself. Many incident plans quietly assume email and Teams will be available to coordinate the response. Agree an out-of-band channel and rehearse it. Write the customer message before you need it. Tell staff today that IT will never ask them to re-enter credentials through a link during an outage. Run a tabletop exercise built on last week’s timeline, since the script has already been written for you.
Resilience budgets get approved in the fortnight after a crisis and quietly cut the quarter after that. Prioritise resilience in the fortnight.


