Distributed Denial of Customer Service

On being helplessly suspended in the cloud.

Nowadays, whenever my site goes down, along with it comes tumbling a significant segment of the consumer web. Yesterday, the Amazon US-East-1 region saw a major outage in their EBS infrastructure, which meant that Instagram, Reddit, Vine, and thousands of other sites went down hard. I don’t know about you, but I can live without sepia photo filters and cute pictures of panda bears for a couple of hours. What I can’t happily accept is that my site went down.

And every developer, company founder, and cloud infrastructure manager around the country felt the same frustration, at precisely the same moment. So what did we all do? We simultaneously emailed, called, and tweeted at our cloud providers—many of which are tragically helpless, as mere abstraction layers atop Amazon’s core infrastructure. My normally hyper-responsive Web n.0 providers all of a sudden choked up, and all they could do was place me in a queue behind the other several thousand requests pouring into their inboxes—in a chorus we all screamed, “I don’t give a damn about those other sites, my site is down!”

On some level I felt bad for the poor people that had to respond to this deluge of frustration, but to be honest, I mostly just wanted them to fix my damn site.

As a community, we all stared blankly at minimalist status pages, impotently refreshing the screen every few seconds, praying for the red status indicator to flip back to green. After nearly two hours, I finally saw that green light saying we were back up—as did thousands of other helpless subscribers. We breathed a collective sigh of relief. We all sent emails around to our bosses and to our customers to tell them we were back, and we all came into the office today, uneasy about the lack of control that comes along with the nebulous abstraction that sits beneath our applications.