Full disclosure. I sold out of my Amazon.Com (AMZN) stock yesterday, and have no conflict in what I'm about to write.
The lessons of this weekend's cloud failures will, in the long run, benefit the biggest, most scaled, and most geographically diverse cloud players, like Amazon.Com and Google (GOOG), in which I retain an interest.
Two things happened.
The most important event was a set of very bad thunderstorms around Washington D.C., impacting Amazon's data center there.
Analysts blamed lazy customers, identifying Netflix (NFLX), Facebook's (FB) Instagram unit (a deal still in the closing process), and some smaller customers for failing to mirror their services in multiple Amazon data centers, which would have kept them running when the Washington center was hit.
Redundancy, in other words, could have prevented the problem from impacting customers and consumers.
The other event that caused problems was wholly the fault of cloud customers. Yelp (YELP), Reddit, and LinkedIn (LNKD) all suffered problems from the addition of a "leap second" at midnight on Saturday, aimed at synchronizing Internet time with the atomic clocks of real time. Those systems with configurations expecting 60 second minutes were knocked down, and although they quickly got back up it was embarrassing.
While there were smarty pants ready to blame all this on "the risks of cloud computing," it wasn't a cloud problem at all, but a data center problem (and a configuration problem). Amazon.com offers its customers the solution of redundancy, but it charges extra for this.
I believe this was a mistake. Redundancy should be a standard offering when you order cloud services. While prices are ferociously competitive among big cloud providers, especially as new entrants like Rackspace (RAX) and Dell (DELL) start calling on customers, making redundancy standard keeps customers from looking stupid and can enhance a cloud provider's reputation enormously.
All this could, in short, prove very bullish for Amazon.Com
Making redundancy standard would also emphasize a key point about the public cloud market, namely that it's a game for scaled, geographically diverse companies only. That means Amazon.Com, Google, IBM (IBM) and only those players able to invest in redundant frameworks will survive the cloud shakeout that's inevitable after the current cloud gold rush ends.