AWS re:Invent 2019 feels like a lifetime ago. So much has happened that's wholly unrelated to the world of AWS -- the pandemic, economic woes, civil strife and more -- that it can be hard to remember everything that occurred at the annual user conference and in the months that followed.
AWS re:Invent used to be a deluge of new products, as the vendor filled gaps in its platform and continued to expand into nearly every conceivable area of IT. Legacy vendors felt the pressure of an aggressive new competitor, while startups were left crestfallen as AWS essentially usurped their niche market.
That's changed over time -- slightly. Cloud computing has matured over the past decade, so the pace of product launches has slowed along with it. At the same time, Amazon has done more to partner with some of the companies it once fought so hard against. Still, it's easy to underestimate just how much Amazon has done to expand its cloud platform over the past year. When you take a step back, the updates cover a wide range of needs.
Before you virtually attend re:Invent 2020, get up to speed and review some of the highlights from last year's show, as well as some important updates to the platform that have happened in the year since.
AWS on premises
The biggest news out of re:Invent 2019 was the general availability of AWS Outposts, an on-premises rack of servers that ran a select number of Amazon cloud services. AWS managed and operated the devices, which sync with the nearest Region for a common interface. It represented the biggest step in a slow but steady change in strategy for AWS, which had long resisted the hybrid cloud model.
Over the course of 2020, AWS has expanded the number of cloud services that could be used on-premises, including support for Amazon Relational Database Service and Amazon S3.
Amazon goes to the edge
Customers' data centers aren't the only places AWS wants to extend its reach. At re:Invent 2019, AWS added Wavelength, which embeds compute and storage services in the data centers of telecommunication providers, such as Verizon. The service aims to capitalize on 5G to support mobile edge computing applications. So far, there are Wavelength Zones in Atlanta, Boston, Dallas, New York, Miami, the San Francisco Bay Area and Washington.
AWS also opened its first Local Zone in Los Angeles. These scaled-back infrastructure offerings act as a logical extension of an AWS Region. They're designed for very low-latency applications, like those for video rendering or virtual desktops. After last year's re:Invent, AWS added a second Local Zone in Los Angeles.
There were smaller devices added to the AWS portfolio as well. AWS Elemental Link is a device that connects live video sources to AWS' media services to process, store and deliver content, either on demand or live. And because there's apparently no shortage of snow-based product names, this summer AWS added Snow Cone to its Snow Family, an 8 TB rugged device for edge computing, storage and data transfers.
AWS also continued to expand its global data center footprint over the past 12 months. News of a new Region or Availability Zone doesn't get the same attention it once did, but it certainly matters to you when one opens closer to your users. Over the past year, AWS added or announced facilities in Canada, India, Italy, Japan, South Africa, South Korea and Switzerland.
Open source expansion
AWS has a complex relationship with the open source community as it remains embroiled in a legal dispute with Elasticsearch over alleged trademark infringement. At the same time, it continues to put out other services based on popular open source projects. At re:Invent, it announced Amazon Managed Apache Cassandra Service, later made generally available as Amazon Keyspaces (for Apache Casandra).
Later in 2020, AWS released Bottlerocket, an open source operating system for container hosting, and put AWS Distro of OpenTelemetry into preview. This observability framework is a project of the Cloud Native Computing Foundation, of which AWS is a member. There was also TorchServe, an open source model server for PyTorch, and the addition of support for RabbitMQ in Amazon MQ, AWS' message broker service.
Isolation and inspection
Amazon continued to extend the ways enterprises can secure their cloud environments. AWS Nitro Enclaves, which was first announced at re:Invent 2019 and recently made generally available, provides isolated Amazon EC2 environments. AWS is taking a slightly different approach than the competition, but the offering is an attractive option for customers who have heightened regulatory requirements.
Amazon Detective added to AWS' growing list of security analysis tools. It processes large amounts of log data to help conduct postmortems on security incidents. This is something administrators could have done manually beforehand, but Amazon Detective automates the necessary steps by incorporating capabilities from Amazon GuardDuty, AWS CloudTrail and Amazon VPC Flow Logs.
And while deeper isolation and analysis help, it's the basics that continue to trip people up. Exposed Amazon S3 buckets remain a problem, despite upgrades in recent years to make them private by default. The latest update to address this problem is the AWS Identity and Access Management (IAM) Access Analyzer.
The tool alerts users when an S3 bucket is publicly accessible. They can then either change the configuration or make other, more granular permission changes in IAM, through a single click.
Simplification through integration and optimization
Simplification was a big theme for AWS at re:Invent 2019 and beyond. Of course, anyone who has spent any time dissecting a large monthly bill from AWS will tell you the cloud vendor hasn't totally cracked the code on simplicity just yet. But there were signs of improvement, such as additional cost management capabilities, for a platform that has been criticized in the past for being overly complex and disjointed.
For Amazon EC2, there was AWS Compute Optimizer, a workload analysis tool that makes recommendations intended to reduce costs and improve performance. There was also the EC2 Image Builder, which automates the creation and management of OS images.
AWS also added several services and tools to simplify app development, troubleshooting and integration. There was Amazon CodeGuru, which identifies an application's most expensive lines of code, and AWS CodeArtifact for software package management. Amazon Kendra added machine learning-backed search capabilities to websites and other interactive applications. And for the less technically savvy, AWS jumped on the no-code bandwagon with Amazon Honeycode.
On the integration side, there was Amazon AppFlow, which moves data securely from SaaS tools to native services. There was also a feature to simplify third-party integrations with an Amazon VPC called VPC Ingress Routing. This tool can automatically route all VPC traffic to an EC2 instance set up for security checks.
And there were tools to help administrators oversee AWS environments, especially for organizations that have many accounts or share data across regions. Amazon S3 Shared Data Sets enables users to easily add applications or teams while maintaining bucket policies. AWS Backup added cross-region capabilities, as did AWS Transit Gateway.
Bigger and faster
AWS' massive scale makes it an excellent candidate for applications that churn through huge data sets. Amazon continues to lean on those capabilities as it expands its portfolio.
Amazon SageMaker, the platform for building machine learning applications, got multiple upgrades in 2020. AWS added an IDE called SageMaker Studio, as well as tools to automate, debug, process, compare, evaluate and monitor models. And Amazon Bracket added a service for building and testing quantum algorithms.
Since AWS engineers are so adept at building and running data centers, it opens the doors to continued performance advancements that trickle down to users. These types of performance upgrades aren't new, but there were some notable improvements at re:Invent and so far in 2020. This includes a growing list of faster and bigger instances that rely on ARM-based Graviton2 processors, and quicker start times for AWS Glue, Amazon Elasticsearch Service and AWS Lambda.