Amazon CTO Werner Vogels took a deep dive Wednesday on the cloud leader's latest capabilities for developers—along the way introducing new data, storage and security services—at the AWS Summit in San Francisco.
But while his keynote focused on new features and best practices for developers to build on the platform, Vogels also noted the important role consulting partners play in driving business to AWS—especially from the large enterprises increasingly going all-in on the cloud.
"We couldn't do it without them," Vogels said of Amazon's channel.
Partners drive enterprise adoption because of the long-term relationships they have with those customers.
"We also see a new breed of system integrators arising that are cloud-native," Vogels said of the evolving channel.
The event in San Francisco largely focused on Amazon's efforts to advance and democratize machine learning.
Matt Wood, AWS general manager of artificial intelligence, in the keynote said the technology is experiencing a renaissance in the cloud, now being used in almost every industry and size of organization.
AWS built a new software stack to drive machine learning development on its platform.
That started with a layer making available several popular frameworks, such as CNTK, MXNet, TensorFlow, Caffe2.
Above that, Amazon built SageMaker, a machine learning platform enabling developers to build, train, tune and deploy models in the cloud through a notebook-style interactive environment.
And for customers looking to entirely avoid development, AWS offers a higher layer of pre-trained application services, such as Rekognition, Polly, Lex, Comprehend, Transcribe and Translate.
Transcribe, the grammatically correct audio-to-text service, and Translate, which translates between languages, both are now generally available, Woods announced.
A new option for SageMaker, called Local Mode, allows the platform to run on developer laptops so they can iterate the process faster by which models are built and trained.
To power those systems, Woods also introduced a new instance, the p3.16xl, that delivers eight Nvidia Volta GPU chips delivering "a petaflop of training performance."
Vogels moved on to discuss new AWS data capabilities and storage options.
Since cloud has made underlying technology available to all, Amazon's CTO said, "what makes our businesses unique now is the data that we have and the quality of the data and how we're operating on that data."
One of the most popular targets of the AWS Database Migration Service is Amazon Aurora—the fastest growing service in the provider's history.
AWS built "a truly modern relational database" in Aurora, meeting customer demand for an "enterprise grade database at open source cost," Vogels said.
Amazon has bulked up its capabilities with Amazon Aurora Multi-Master, which Vogels described as the first relational database that can scale out across multiple data centers.
And to help customers deal with unpredictable workloads, Amazon Aurora Serverless allows administrators to ignore the underlying server nodes running their databases by charging only for capacity used.
The service "scales up and down depending on the traffic that's coming to your database," Vogels said. It "brings relational databases into the true serverless cloud fashion."
Vogels introduced a new object storage class: S3 One-Zone Infrequent Access.
The current offering, S3 Standard-Infrequent Access, is used for storing data that doesn't need to be touched very often, but needs faster access speeds than cold storage.
The new option implements that model with even less availability—only two nines of guaranteed uptime—because it runs in only one availability zone. But customers that can live with that downgrade will save another 20 percent.
Vogels closed his keynote by focusing on security services that help implement the best practices Amazon describes through its AWS Well-Architected Framework.
Developers should be building applications "with security by design as a first principle," he said.
The security team shouldn't come in at the end of the development process. It should be "integrated into everything in our lives and in our development," Vogels said.
That philosophy is behind the launch of AWS Secrets Manager, which makes it "so you never ever again have to put a secret in your code."
The service enables customers to rotate and manage access to credentials and keys for AWS and third-party services, as well as on-premises systems.
"It allows us to build systems that are way more secure than we could ever do in the past," he said.
"We can do automatic database password rotation for you. You never have to look at a database password ever again."
He also announced the launch of a new data-aggregation feature in AWS Config that centralizes auditing and governance capabilities.
Another new feature, AWS Firewall manager, helps customers manage web firewall rules over multiple accounts and multiple applications.
To make encryption more ubiquitous, especially personal information of customers and critical business data, AWS is launching the AWS Certificate Manager, a private version of its certificate authority that works on services accessible only internally in organizations.
Amazon also will extend its Fargate service, which abstracts containers away from their underlying virtual machine infrastructure, to its managed Kubernetes service sometime soon with AWS Fargate for EKS, said Abby Fuller, senior technical evangelist at AWS.