top of page

The NSA's Classified Network Went Down Today, Reportedly After Patching. Staff Were Told to Stagger Arrivals So Everyone Didn't Log In at Once. That's the Fourth Outage in Six Days

Writer: Patrick Duggan
Patrick Duggan
52 minutes ago
4 min read

If you run IT anywhere, take the one concrete detail from today's NSA outage and put it in your own recovery plan: when a system comes back, don't let everyone reconnect at the same moment. According to CBS News sources, the NSA staggered arrival times at Fort Meade so large numbers of employees would not log on at once while its network was struggling. Microsoft learned the same lesson the hard way last Thursday. Write down now how you would bring users back in waves.



What is reported


On Tuesday, November 17, 2026, the NSA's classified internal network went down. CNN broke the story, and CBS News and Nextgov/FCW followed, citing their own sources.


  • The network. CBS calls it "NSA Net," used "for classified communications and information sharing" inside the agency. Nextgov calls it "the primary platform for transmitting top-secret data."

  • The cause, per anonymous sources. Two CBS sources said it "resulted from routine system patching, though the precise circumstances surrounding the disruption remain unclear." Four of Nextgov's six sources said it "may be linked to a series of software upgrades." A U.S. official quoted by CBS, not identified as NSA, called it a "temporary degradation during scheduled routine maintenance."

  • Not an attack, so far. Two U.S. officials told CBS there is "no evidence the disruption was caused by a cyberattack." CNN and Nextgov report the same from their sources.

  • The workaround. Per CBS sources, non-emergency staff got staggered arrival times, and "some activities are being redirected to other NSA facilities around the country." The official told CBS that "mission/critical capabilities remain fully operational."

  • The President. CNN's sources said the outage would not affect urgent intelligence reaching the President, citing "multiple redundancies and backup networks."

Everything above comes from people speaking anonymously or on background. No NSA, ODNI or CISA statement on the record had appeared when we wrote this. Nobody has said when full service will return.



Why the staggered arrivals matter


Last Thursday, Microsoft 365 went down worldwide. Microsoft's explanation, as quoted by ABC News, includes this line: as servers came back, "they were immediately overwhelmed by waiting traffic and became unresponsive again." Every user and every app retried at once, and the recovering servers fell over again.


The NSA appears to have done by hand what software should do automatically: let people back in slowly. Telling thousands of people to arrive in waves is admission control with a parking lot. It works, and it is also a sign that the system could not protect itself from its own users coming back.


For your own systems, that means:


  1. Bring users back in waves. Restore access by group, site or shift, not everyone at 8 a.m.

  2. Make clients back off. Apps that retry instantly and forever turn a recovery into a second outage. Retries should wait longer each time, with some randomness.

  3. Have a quiet way to say "not yet." A status page or text message that tells people when their wave starts keeps them from hammering the login page to find out.


Four in six days



Day

System

Cause, as stated so far

Thu Nov 12

Microsoft 365

A latent bug from a July fix, a traffic surge, and reduced capacity

Fri Nov 13

FAA air traffic telecom

Data flooding from two remote sites, one mid router replacement; the FAA disabled the broadcast function on the new routers

Mon Nov 16

Chase and the banks connected to it through Zelle

A software problem at Chase, per Chase via the press

Tue Nov 17

NSA Net

Patching or software upgrades, per anonymous sources


None of the four is reported as an attack. All four are, as far as anyone has said, ordinary maintenance that went wrong. Nextgov made the same connection, noting that "routine technical issues have been linked as the cause of these previous events."


We wrote up the pattern behind the first three, and the five questions to ask before any change, in Forgot about 2009. The NSA outage fits it: a change, a system without enough margin to absorb it, and a recovery that had to be managed so it wouldn't knock itself over again.



What to do this week


  • Before your next patch window: check that you can roll back, that someone is watching while it happens, and that you know how users will reconnect afterwards.

  • Don't freeze all patching because of this. The answer to bad changes is better change control, not no changes. Unpatched systems are how attackers get in, and the NSA's own advisories say so constantly.

  • Read official word when it arrives. If the NSA or ODNI publishes a cause, it replaces the anonymous accounts above, and we will update this post.


What is not known


  • What was patched or upgraded, and how it broke the network.

  • How long the network was down, and whether it is back.

  • Whether the staggered arrivals were about load, as CBS's sources describe, or about something else.


Sources


CBS News, "NSA's classified internal network goes down, sources say," November 17, 2026. Nextgov/FCW, "Classified NSA communications network down, people familiar say," November 17, 2026. CNN's original report, read via WRAL's republication of the CNN wire, November 17, 2026 (cnn.com could not be read directly). ABC News on Microsoft's November 12 outage explanation, November 16, 2026. FAA, "Nationwide Ground Stop: Questions and Answers," November 15, 2026.




How do AI models see YOUR brand?

AIPM has audited 250+ domains. 15 seconds. Free while still in beta.


Comments

Rated 0 out of 5 stars.
No ratings yet

Add a rating
bottom of page