If you overcook the roast, it’s not the oven’s fault. That’s one way to think about Amazon Web Services’ connection to the embarrassing Prime Day glitches that marred the ordering and product-search features on Amazon.com during one of its biggest days of the year.
Following the widely publicized issues, it might be tempting to think that they could cause serious problems for AWS, which counts Amazon as its first and most-important customer. But that would be misleading: AWS itself had a pretty normal day Monday, with nearly every service running as expected on its status dashboard. Some people did have problems logging into the AWS Management Console, but that is a tool that allows customers to manage things like account spending records and security controls and has zero impact on the performance of applications running on AWS.
AWS issued this statement Tuesday: “AWS continues to function normally. We saw some intermittent AWS Management Console issues (yesterday), but they did not drive any meaningful impact on Amazon’s Consumer Business.” While the Amazon side of the house did not immediately respond to a request for more information about what actually did happen, a source familiar with the events Monday said that the Prime Day glitches were not connected to anything on the AWS end.
It’s easy to think of Amazon and AWS as one big conglomerate, but AWS CEO Andy Jassy often describes the cloud division as “a separable business” that treats Amazon’s other businesses the same way they would any other big customer. And in cloud computing, customers are responsible for the development, testing, and deployment of their applications on cloud services. If you write something buggy, and it doesn’t work correctly at scale, that’s not the cloud provider’s fault any more than it would be if you were managing your own infrastructure and something broke in your application code; you’re not going to call up Intel and complain.
Cloud buyers know this. They turn to cloud infrastructure services to remove the complexity of managing hardware; of having to pay someone to walk around data centers swapping out hard drives as they fail, or managing the power supplies to those buildings. Software development is still the responsibility of the customer, and while there is a lot cloud providers can do to help make everything go smoothly, if their services are running as expected and something breaks with your software, that is your problem.
“I don’t think this incident affects the reputation of the underlying cloud services provider,” said Fernando Montenegro, an analyst with 451 Research. “AWS and other cloud providers are quite serious about availability, and provide lots of guidance for their customers to build highly resilient applications on top of the underlying services.”
So while it might be fun to snark on Twitter about it, anyone who makes a cloud buying decision based on Amazon’s performance during Prime Day hasn’t really thought it through. If you’re wondering what might slow down the cloud giant, focus on the march of the retail industry to rival cloud providers, as evidenced by the huge Microsoft-Walmart deal signed Monday night.