AWS re:Invent Conference Shifts Focus From VMs To Containers and Serverless Computing

A day after AWS introduced several new virtual machines to partners, Amazon CTO Werner Vogels shifted focus. In the last keynote of the public cloud leader's re:Invent conference on Thursday, the big topic was container-based and serverless computing models.

Taken together, the aim of these new computing models is to help solution providers become more efficient in how they provide services to enterprises, allowing them to spend less time on infrastructure tasks and more time creating customer value.

Modern approaches to deploying cloud applications, sans VM instances, have been a major emphasis at AWS over the last year. The company has made them available to partners through the EC2 Container Service (ECS) and the Lambda function that executes basic code.

Vogels revealed several updates to make those services more comprehensive and secure. He also introduced some tools that excited the developers in the audience, as well as data processing and operations capabilities.

[Related: AWS Delivers More AI, Analytics, Partnerships And A Big Rig For Moving Big Data]

"Virtual machines are very important, and I saw in Andy's keynote a set of new instance types you can be using," Vogels said, referring to AWS CEO Andy Jassy's announcements on Wednesday.

But there's a spectrum of approaches to implementing compute environments, fulfilling different needs, that spans VMs to containers to serverless functions, Vogels said.

The proliferation of microservices is driving widespread container adoption, and using the open source products associated with the technology "is really a pain," Vogels said. "You have to go outside the cloud world to do these things."

ECS "is really is the tool you want to use for all of this," Vogels told attendees, because that service delivers the power of cloud when orchestrating and managing container clusters.

To advance container proficiency in the cloud, Vogels introduced a new open source community that Amazon will foster. That community, Blox, represents a collection of open source projects for implementing containers in ECS.

The first two projects Blox is tackling involve a cluster-state service and a daemon-scheduler.

Since the introduction of Lambda at last year's re:Invent, serverless computing is rapidly gaining popularity in the AWS ecosystem.

The serverless approach isn't just for small companies, he said.

"Many large enterprises have figured out that this is the way that's most cost-effective for them to do large execution," Vogels said.

But it's important to remember that the innovative paradigm for executing code independent of VM instances involves more than just Lambda.

"Serverless is the whole collection of meta-services that we give you," Vogels said, from databases to data ingestion engines to analytics tools. "Lambda is just the execution engine."

Beefing up that function, AWS added C# to the list of Lambda-compatible languages.

Vogels also introduced AWS Lambda@Edge, which allows Lambda functions to be executed at Amazon's CDN locations on the edge of networks, rather than in the cloud data center.

"Coordinating very large scale Lambda functions is a pain," Vogels said "How do you step through such an endless execution environment to make sure everything executes efficiently?"

To that end, he announced the general availability of AWS Step Functions, a service for coordinating different components of a Lambda function through state machines.

Mike Kavis, vice president and principal architect at Cloud Technology Partners, an AWS systems integration partner based in Boston, told CRN the serverless technologies are part of the evolution of the cloud, where "everything's about getting faster."

As the cloud matures, capabilities that once were such an upgrade in speed, like provisioning VMs, now seem slow to users, who just want to execute applications.

"I'm not still mucking around with infrastructure, mucking around with the operating system, I'm writing code," Kavis said. AWS is "escalating services to a higher level, so what's underneath them becomes less relevant."

Containers and Lambda are "getting people working on higher-value things," Kavis said.

In many ways, those services are similar to Platform-as-a-Service, in that they add another layer of abstraction away from infrastructure, Kavis told CRN. "The more you're touching infrastructure, the less value you're providing to the customer."

Before talking about containers and serverless computing, Vogels delved into many tools for developers, operations professionals and data engineers.

On the operations front, Vogels introduced AWS OpsWorks For Chef, which offers a fully automated server running Chef, the market's leading configuration manager. The managed service solves "one of the smaller but very important pain points" for AWS customers who use Chef cookbooks to configure their environments, he said.

Amazon EC2 Systems Manager is a new collection of tools that packages installation, patching resource configuration and task automation.

The new Amazon CodeBuild completes a trilogy of services—with CodeCommit and CodeDeploy—to scale a continuous development pipeline.

"Most of our customers who want to move fast go to a continuous integration and continuous delivery model," Vogels said, but it can be quite difficult to check and secure 15,000 lines of code.

The CI/CD solution is metered by the minute and automatically scales.

To boost monitoring and visibility, Vogels introduced AWS X-Ray, a free managed service which allows developers to look into their applications and service execution.

AWS Personal Health Dashboard delivers to customers a view of system events for all their workloads across all regions. The dashboard will automatically notify users of failovers and can automate responses to system events by executing Lambda functions.

On the security front, AWS released a solution to guard against Distributed Denial of Service (DDoS) attacks.

AWS Shield For Everyone—turned on by default—protects customers against the volumetric attacks that "flood the pipe" and more sophisticated "state-exhaustion attacks."

AWS Shield Advanced gives customers 24/7 access to a DDoS response and support team to protect against larger and more sophisticated attacks against applications.

Data storage and management was another major area of emphasis throughout Vogel's keynote.

A new service called S3 Storage Management adds to Amazon's popular storage service tagging, analysis, metrics and inventory capabilities.

AWS Glue is a fully-managed data catalog that pieces together data management features.

"With AWS Glue and all other AWS services you have available to you, you're now able to build a comprehensive data architecture on AWS," Vogels said.

AWS Batch gives developers the ability to dynamically provision compute resources and optimize their distribution, solving a common pain point for large-scale batch processing, Vogels said.