Maybe big businesses really do understand cloud computing after all.
When I sat down with Amazon CTO Werner Vogels at the Amazon Web Services re:Invent conference on Wednesday, we began the discussion by talking about applications designed to take advantage of everything the cloud has to offer in terms of control, resiliency and programmability — what Vogels calls 21st-century architectures. It’s great in theory but, I asked, “Who’s actually building these apps?”
Mainly, I was concerned with whether AWS has been able to prod the enterprise customers it so desperately desires into adopting Vogels’s design principles for their applications. Often times, any discussion about “enterprise cloud computing” begins and ends with whether they can run their legacy SAP applications on cloud servers.
Vogels acknowledged there are plenty of reasons to focus on those legacy applications, including the decreased costs and increased flexibility the cloud can bring. He pointed to human resources applications as one of the first that moved to the cloud, citing Amazon’s own internal employee-review process that requires some 60,000 employees to file performance reviews on March 1 every year. He calls these “IT life events” that get companies thinking seriously about whether it’s worth their time and money to invest in more hardware.
Scaling from 1 server to 200 servers for just a week or so is a way to achieve serious return on investment “without radically innovating in terms of software,” Vogels said. However, while that’s a topic rife with financial implications, it doesn’t address the spirit of cloud-enabled innovation that companies such as AWS tout so loudly.
Just consider what’s possible. Anyone who follows cloud computing is familiar with how Netflix has a built a service-oriented, highly dynamic cloud infrastructure atop AWS that’s built to withstand whatever service disruptions come its way. Vogels and I also discussed Pinterest and its cloud system, which is designed to maximize cost savings by relying heavily on Reserved Instances for the base load and then programmatically favoring Spot Instances over On-Demand Instances when more resources are required. (“These guys have turned [cloud pricing] completely on its head,” Vogels said.)
However, he added, although AWS uses startups to illustrate cool architectures, “we also could invite Samsung on stage.” Samsung’s smart TVs get their intelligence from software, and all that runs in the AWS cloud. “That architecture is not an 19th-century architecture,” Vogels said. “That architecture is a 21st-century architecture.”
Or take media companies, where Vogels said many traditionally print media companies are looking for new ways to monetize their content. The United Kingdom’s Telegraph used AWS to power an application that makes it easy for readers to buy what they see on the website’s fashion section, he explained. That attracted more advertising dollars, he said, and could be expanded into coverage about cars, music and other consumer-centric content.
ABC’s mobile app for watching live content splits uploads into parallel streams, does live transcoding and automatically inserts advertisements based on where viewers are geographically located at any given time. Other media companies are using advanced analytics running in AWS’s cloud to figure out what content viewers want to read and watch on their mobile devices, and how they want to consume it.
Shell Oil is using the cloud for continuous deployment and testing of applications, and pharmaceutical companies are for the first time ever doing true collaborative research thanks to the cloud. All sorts of smart devices use the cloud as the backend for storing and processing data.
“Are these old-style IT architectures?” Vogels asked. “No.”
They are, however, new applications that aren’t tied to lucrative enterprise software markets built over decades. Rather than the mission-critical applications that many cloud providers — including AWS — are trying to convince enterprise users to migrate to the cloud, these new applications are relatively inexpensive to build and run. And, Vogels, noted, if these applications don’t work, companies can pull the plug on them with relatively minimal business impact.
He sees more companies coming around on this type of innovation through applications. “Maybe sometimes experimentation is a hard word, it’s like doing research or science,” Vogels said, but when companies are facing increasing consumer choices and decreasing customer loyalty, they have to be agile and figure out what works in order to keep their businesses growing.