Pricing structure isn't the only concern, though. There are perceived risks around security, certifications, and privacy, which is why so many of the AWS use cases start as small test and development projects with non-sensitive data. "Shadow IT," when employees or business units within an organization use IT resources like AWS cloud services without informing their IT department, is an entry point for many organizations that lead to more formalized uses of AWS resources. "As soon as those cloud projects start producing revenue, or start impacting the bottom line, that's when IT gets involved," says Peder Ulander of Citrix, whose cloud management platform integrates with AWS services.
AWS says despite many enterprises not being willing to jump fully into the public cloud, adoption has "accelerated rapidly" in the enterprise market, says Adam Selipsky, VP of marketing, sales and support for AWS.
Jeff Bezos, the founder of AWS's partner company Amazon.com, said when asked that what surprised him most about AWS has been the adoption by enterprises, government, and education institutions despite the service being only six and a half years old.
"For enterprises it's a process," to move to the cloud, Selipsky says, adding that even Amazon.com, one of AWS's biggest users, has taken a multi-year migrations strategy to using cloud resources. "We think a significant majority of workloads will move to the cloud," he says.
SHOWDOWN: 15 Hottest Tech Arguments
Gartner analyst Lydia Leong says many enterprise deployments of AWS services are still driven by specific use cases. There are notable exceptions, such as Netflix, one of AWS's poster customers, which runs almost its entire business on Amazon's cloud. But a more common enterprise use case is for public cloud to run IT workloads beyond the scope of their internal resources. Perhaps no company better exemplifies that than Cycle Computing.
CEO Jason Stowe helps companies use AWS resources for high-performance computing needs. Cycle works with a majority of the top 20 pharmaceutical companies, he says, including Novartis, which ran a 30,000-core workload over 95,000 compute hours in AWS's cloud to run analysis of drug tests. It's a game-changer for the enterprise, he contends.
"This is like moving from the horse and buggy to the automobile," he says. As Cycle Computing ramps up, he says large-scale workloads of 10,000 to 20,000 cores have becoming "pedestrian" for Cycle to manage and AWS to handle. Robert Half International, The Hartford, and Pacific Life all discussed at Amazon's conference ways in which they're doing HPC in AWS's cloud.
Still, when Novartis, Pacific Life, and other major enterprises are doing these massive workloads on AWS's cloud, it's not the company's entire IT operations running the cloud; it's a couple of really big workloads.
"The public cloud is great for that spike of usage and resources that are needed," says Zev Lederman, of Newvem, which helps companies track and optimize their AWS usage. For AWS, and the public cloud in general, to move from intermittent use to consistent, all-the-time usage, customers, he says, need to have better tools to manage their cloud workloads, which is what Newvem helps companies do.