The day Google's Marissa Mayer "broke the internet"

"It was 8:27 a.m. on a Saturday in 2009. I was in Minneapolis visiting my brother. I woke up late and was racing around my hotel room trying to get ready. There was a knock on the door. I was sure it was my mom, telling me I was late. But it was my friend Jini. 'Yes, you're late to breakfast,' she said calmly. 'But also: Google is down. I think you might want to deal with that first."—Google's Marissa Mayer on the "slash heard 'round the world," in Newsweek.

• [adonai]

  The day Google’s Marissa Mayer “broke the internet”

  You mean she typed “Google” into Google?

• [cory]

  Shouldn’t this be titled “the day Marissa Mayer fixed the Internet” or something? No indication it was actually her fault.

• [shutz]

  This reminds me of 50-word story I wrote a few years ago (from what I can determine, this was probably before the “slash” incident.)

  The End of the World, As We Know It

  “Oh shit!” I heard her scream from her room.

  Then she ran past my room, saying, “Grab your stuff. Only the important stuff. No, leave your laptop.”

  Perplexed, I asked, “What’s going on?”

  “The end of the world! The shit has hit the fan!”

  “What?”

  “Google is going down. Permanently.”

  You can find this story, and more like it, here.

• [sn00py]

  What’s impressive is not that they found and fixed the problem “for a system as complex as Google” in such a short time. It’s that the Google infrastructure is evidently so well designed that such rapid troubleshooting is possible. I’m not trivializing what she did, being a seasoned firefighter myself. But finding a single stray character could have taken many, many excruciating hours if Google were a poorly designed rubber band ball. And they’re lucky it wasn’t a bug in a third-party proprietary package.

  • [Ocker3]

    Agree. Without proper dynamic load-tracking apps, visibility of where data was going, where it wasn’t going, etc., this would have been a days problem, not a minutes (60) problem. In this case, all the time/money they spent on insurance (analytical systems of their own systems) really paid off. The longer I work in IT, the more I believe in buying smart Insurance/service policies, no matter how high the manufacturing standard. An insurance policy for a well-manufactured/designed product should of course be cheaper than for other products, but there are never any solid guaruntees that something Won’t Ever need replacement/maintenance.

• [Joshua Ochs]

  There are two distinct possibilities here:
  1) Amazing analytics tracked down the problem immediately.
  2) They called up the team of 3 or so people who manage the safety blacklist, and said “Who changed something in the last 6 hours? Oh, it was Bob? What did he do? Yeah – undo that.”
  I’m betting on the latter over fantasy analytics, even for Google.

  This is also why I criticized the article the way I did – there is absolutely no detail, no information here. It’s “there was a problem, quickly fixed”. That’s daily life, folks, not a news article.

• [Joshua Ochs]

  Article summary:

  “I woke up.
  Something broke?
  Oh hey, we’re huge!
  We fixed it.
  I ate breakfast.”

  (Breakfast not necessarily included in article, see your terms and conditions for details.)