AWS re:Invent 2018 another test for the cloud giant

AWS hybrid strategy

For years, AWS publicly urged enterprises to go all-in on its cloud and dismissed hybrid cloud as counterproductive to future success. But despite its steadfast rhetoric, AWS has ceded to the demands of enterprise clients that have dozens, if not hundreds, of applications that are difficult to migrate.

"They talked about false cloud and fake cloud and used language like that to position public cloud very clearly," said Paul Miller, a Forrester Research analyst. "[But] behind the scenes, they understand the reality for most customers is hybrid."

It's an acknowledgement that there is interest in higher-layer [AWS] services, but maybe the way it's delivered is not through EC2 environments.

That hybrid strategy came to a head in 2017 with the rollout of VMware Cloud on AWS, a service that enables enterprises to port the familiar vSphere environment directly to Amazon's cloud. Recently added services extend some of AWS' software beyond its own network and into edge devices and private data centers.

Industry observers expect that strategy to go even further at AWS re:Invent 2018, particularly in light of this past August's public preview of an on-premises version of Amazon Relational Database Service (RDS) in conjunction with VMware.

"It's a completely new direction for them," said Deepak Mohan, a research director at IDC. "It's an acknowledgement that there is interest in higher-layer [AWS] services, but maybe the way it's delivered is not through EC2 environments."

Amazon RDS on VMware is essentially a SaaS version of AWS' database service, Mohan said, and could open the door to bring more of Amazon's services, such as RedShift, Aurora or its machine learning offerings, on premises.

Behind the scenes, they understand the reality for most customers is hybrid.

Microsoft Azure's hybrid strategy represents one of the biggest long-term threats to AWS, said Rhett Dillingham, an analyst at Moor Insights & Strategy. Amazon RDS on VMware is a big step in the right direction, but it still relies on the VMware stack and what AWS can add on top.

"If the enterprise is comfortable with the VMware software stack and is running that across private infrastructure and AWS, then there is no perceived AWS gap in hybrid," he said. "But, for an enterprise that is not looking to extend its VMware footprint into public cloud, they see a gap versus Azure's hybrid solution."

Managed services

AWS has built a massive ecosystem of partners to help enterprises manage the platform and to link its cloud to other prominent technologies. It's often squeezed out smaller companies when it wants to directly offer a competing service or feature -- so much so that the re:Invent keynotes have earned a reputation as a death knell for many third-party vendors in attendance. But AWS also relies heavily on that network of companies to help users navigate a sprawling set of services and a delivery model that carries a steep learning curve for traditional IT shops.

There are always questions about what that balance will look like coming out of the AWS re:Invent 2018 conference. In recent years, AWS has added more managed services, but it's also done more to support third parties and even carved out a stand-alone partner summit as part of the weeklong conference. Still, industry observers expect -- or at least hope -- to see AWS be more hands on.

"They've always positioned it as, 'We've got this fabulous platform, and we've got partners that will help you integrate into it,'" said Melanie Posey, a 451 Research analyst. A new strategy, she said, may call for AWS to become more involved in integrations and professional services.

Posey compared the arc of cloud adoption to a baseball season, saying it's just getting to the late stages of spring training. To meet the needs of those enterprises, AWS must offer a more finished product, she said.

There's also interest in greater ease of consumption. AWS has added features at a daunting pace for more than a decade, so it's probably a good time for the company to tidy up the catalog and make it easier to guide users through their options, Miller said.

"Even for someone who is deeply expert in one area and knows all the options inside and out, they're going to get overwhelmed when they get to another piece of the puzzle," he said.

Users have welcomed many of the services AWS has added over the years, but there are concerns that it's done so at the expense of functionality for some of its core products.

CloudFormation's GUI looks like it was designed 15 years ago, said Brian Tarbox, lead cloud engineer at Cogito and a SearchAWS contributor. For example, users can view nested stacks but can't break that down to show only top-level stacks.

"Just some of the basics," Tarbox said. "I understand they are going a mile a minute, and, in most places, they have done a remarkable job with all the cross-integration. But, it's like, guys, slow down and put just a little bit of work into your GUIs."

Machine learning

One answer to the deluge of services and management responsibilities could be AI. AWS has already moved in this direction in security with its GuardDuty and Macie services, which use machine learning to detect threats and ensure proper security policies are followed. AWS will likely double down on the use of AI on the back end to help users, possibly in the areas of cost optimization and general operational health, said Jason McKay, CTO at Logicworks, an AWS managed services provider.

There's also an expectation that AWS will continue to bolster its customer-facing AI tools -- from its higher-level cognitive services to SageMaker, which makes machine learning more approachable on its platform for data scientists and analysts.

TechTarget Senior Site Editor Kristin Knapp contributed to this report.