
Two major outages in two days raised the same question I’d ask in my own operations center… What changed right before things failed?
On Thursday, November 12, Microsoft 365 started failing. Outlook, Teams, and other services people rely on to run their businesses were disrupted. I spent part of Friday dealing with the fallout from the IT side. Then, while Microsoft was still recovering, the FAA halted departures nationwide for more than four hours after problems with radar and communications. apnews.com
The timing got my attention. The explanations kept it. In both cases, capacity and infrastructure became central to the public account. And in both cases, I came away asking the same thing… What happened first?
Microsoft… What took capacity out of service?
Microsoft said a problem introduced during an earlier network capacity upgrade surfaced after a subsequent network change. In its Exchange Online updates, it also said some servers became unexpectedly unavailable during automated checks. Microsoft said it found no evidence of malicious activity and committed to releasing a full post-incident report. apnews.com
That gives us a sequence worth examining. What did the later network change do? Did the Exchange servers stop functioning, or did the automated checks lose the ability to reach them? Where did their workload go? How much capacity was available to absorb it?
If the remaining infrastructure couldn’t handle the load, that could explain how the outage spread. I still want to see what caused the first loss of capacity and why failover couldn’t contain it. A post-incident report should connect those events with timestamps and telemetry. That’s how customers can judge what failed and whether the proposed fixes address it.
The FAA… The equipment was old on Thursday, too
Transportation Secretary Sean Duffy said legacy switches and routers became overloaded and started producing inconsistent data. Officials also linked the overload to new technology being added to the air traffic control system. Homeland Security Secretary Markwayne Mullin said there was no indication of a cyberattack at that point. rollcall.com
I can believe that aging equipment had limited capacity. It had limited capacity the day before, too. What put it over the limit on Friday? Did the new technology change the volume or pattern of traffic? Did another failure move traffic onto equipment that was already close to capacity? What did the network telemetry show in the minutes before controllers began getting inconsistent information?
That leads directly to testing. I haven’t been able to verify the specific claim that officials said they should have tested with a smaller group, but I’ve been listening to that additional explantion on some live news reports. What officials are now saying about new technology being added to the live system is enough to ask how the change was tested and introduced. abcnews.com
In a system this critical, I’d expect the new and legacy equipment to be tested together under representative load before a live rollout. I’d also expect the team to test what happens when a shared component fails or starts returning inconsistent data. A limited production pilot can reveal conditions a test environment can’t reproduce. But a pilot is only as contained as the infrastructure it depends on. A handful of facilities can still share a network component with the rest of the system.
So what passed in pre-production? Which facilities and shared systems were exposed to the live change? What conditions would have stopped the rollout, and how quickly could the team have backed it out? If my team’s explanation after an outage was “we should have started with a smaller group,” those would be my next questions.
What makes Thanksgiving different?
Duffy said he expected operations to be back up for peak Thanksgiving travel. By then, the FAA had already lifted the ground stop. Restoring service in hours on existing equipment is plausible, even when replacing that equipment takes years. My question is what supports the confidence about peak travel. rollcall.com
What did the FAA change to stabilize operations? What capacity does the system have now? Has the FAA verified that the conditions that triggered Friday’s failure can be handled if they recur during the holiday rush? Getting flights moving again tells me the FAA restored operations. The answers to those questions would tell me how much confidence to place in the forecast.
What did “no evidence of a cyberattack” mean at the time?
After years in IT and security operations, I’ve learned to pay attention to when that statement is made and what the team has reviewed by then. During an active outage, people are trying to restore service while investigators reconstruct what happened. A preliminary statement can be accurate at the time it’s made. Its weight depends on the evidence behind it.
If two incidents of this scale happened that close together on my watch, I’d compare their timelines and look for shared dependencies or matching anomalies before either failure. Officials said they saw no indication that the FAA outage was related to Microsoft’s. I’d like to know what they compared. The timing is reason to ask. The public record so far gives me no affirmative evidence of a cyberattack or a shared cause. www.nextgov.com
What I call BS on is treating “the infrastructure couldn’t handle the load” as a finished explanation. If my team brought me that as a root cause, I’d send it back. Show me what changed, what failed first, and what you checked before telling people you saw no indication of an attack. I’d hold Microsoft and the FAA to the same standard.





