AWS re:Invent 2021: The biggest announcements
AWS re:Invent 2021 was packed with announcements. Our team of gurus narrows down the news to highlight the top stories from this year's event. Read on!
Jun 08, 2023 • 11 Minute Read
As AWS re:Invent 2021 comes to a close, cloud engineers have a plethora of new tools in their cloud toolboxes. Missed the event or looking for the TL;DR on what AWS announced? Here are our team's picks of the biggest announcements that came out of re:Invent 2021.
Accelerate your career
Get started with ACG and transform your career with courses and real hands-on labs in AWS, Microsoft Azure, Google Cloud, and beyond.
AWS Private 5G lets you build your own private cell network
AWS Private 5G will let you build your own private 5G cell network. This service, which is in preview in the US, could be ideal for uses cases like a warehouse or large campus where you’re looking to help increase bandwidth and coverage for an increasing number of devices while maintain the security and control of a private network.
AWS ships you everything you need, including SIM cards. And there are no upfront or per-device costs. Customers just pay for the network capacity and throughput they request.
SageMaker Canvas offers no-code ML model building
There was a lot of talk around data and machine learning this year, starting with SageMaker, which had several new announcements. Starting with...
SageMaker Canvas is now in GA. Canvas is designed to help business users generate ML predictions using a visual, no-code interface. This allows people like business analysts who may have little or no ML experience to get in and get valuable insights out of data.
SageMaker Ground Truth Plus aims to simplify ML scaling
SageMaker Ground Truth Plus, is also in GA. This new turnkey service lets you tap into a team of experts to create training datasets without having to build and manage labeling applications on your own.
You provide the data and labeling requirements and your data labeling workflows get set up and managed for you. Amazon claims this can reduces your costs by up to 40%. Not bad!
Amazon SageMaker Serverless Inference is in preview
AWS announced Amazon SageMaker Serverless Inference. This lets you easily deploy machine learning models for inference without the need to configure or manage underlying infrastructure.
This is ideal for applications with intermittent or unpredicatable traffic — you only pay for the duration of running the inference code and the amount of data processed, not for idle time.
SageMaker Studio Lab to help ML learners
AWS Announced SageMaker Studio Lab, which now in preview. This free service is designed to help anyone learn and experiment with ML without the need for an AWS account, credit card, or cloud configuration knowledge.
The ML learning curve can be a steep one, so with a predictable and controlled environment for hosting your Jupyter notebooks, you can ensure you won’t be racking up huge bills while you’re trying to learn.
Trn1 chips for ML models
Moving on from SageMaker, AWS announced that AWS Trainium-based Amazon EC2 Trn1 instances are in preview. These instances are optimized for training deep-learning models in the cloud and offer best price-performance for model training and 800 Gbps networking bandwidth. These are ideal for use cases like language processing or image recognition.
Amazon DevOps Guru for RDS helps with Aurora-related issues
Amazon announced Amazon DevOps Guru for RDS. This new capability for DevOps Guru makes it easier for developers to detect, diagnose, and resolve issues in Aurora, and it’s available in all Regions where DevOps Guru is available.
Yes, AWS knows it has a lot of services (but here are a few more)
For perspective on how much AWS has grown (not just in 15 years, but in the last single year), consider the number of services available. And it just got bigger…
New Graviton3 processors to power C7g instances
Amazon announced its next-generation of ARM-based Graivton3 processors, boasting 25% better performance and 60% less power usage. These chips will power the new Amazon EC2 C7g Instances, which are now in preview and are ideal for compute-intensive workloads, like high-performance computing, scientific modelling, and CPU-based machine learning inferencing.
EC2 M1 Mac instances announced for building Apple apps
Last year, Amazon announced EC2 Mac instances. This year, they announced the preview of new EC2 M1 Mac Instances. This upgrade includes Apple’s latest and greatest M1 chip and will be great for developers building for iPhone, iPad, Apple Watch, and Apple TV.
AWS Amplify Studio offers low-code development
We’ve seen several low- and no-code services announced this week. Add to that list the announcement of AWS Amplify Studio. This service helps you visually build feature-rich applications in hours instead of weeks using a visual interface that requires little knowledge of how to write code. Amplify Studio will allow developers to set up a backend, create UI components, and connect them together. The service is currently in preview.
AWS adds serverless and on-demand analytics options
AWS announced serverless and on-demand options are now available for Redshift, EMR, MSK, and Kinesis in preview. This means no configuration or scaling of clusters or servers. Just fire them up and the services scale up when you’re busy and scale down when you’re not. You pay only when service is in use.
AWS introduces open-source Kubernetes autoscaler Karpenter
To help simplify Kubernetes infrastructure, AWS has announced Karpenter — with a “K” … because, you know, Kubernetes. This open-source autoscaling tool is aimed at improving application availability and cluster efficiency. Karpenter automatically launches needed compute resources as application loads change. This should be welcome news to many as AWS says about half of its Kubernetes customers report Kubernetes Cluster Autoscaler is challenging and restrictive.
An avalanche of S3 updates
But one of the announcement that stands out is the new Amazon S3 Glacier Instant Retrieval storage class. This new class will give the lowest cost storage for data that’s rarely accessed (up to 68% cheaper than S3 Standard-Infrequent Access) while offering milliseconds retrieval — perfect for use cases like media archives or medical images.
The existing Amazon S3 Glacier storage class has now been renamed S3 Glacier Flexible Retrieval and offers free bulk retrievals in 5 to 12 hours, and the storage price has been reduced by 10 percent in all Regions!
New DynamoDB Standard-Infrequent Access table class
How does saving 60% on DynamoDB costs sound? Well, the new Standard-Infrequent Access table class for DynamoDB aims to let you do just that.
DynamoDB Standard-IA is Ideal for long-term storage data that is infrequently accessed, such as application logs or e-commerce order history. This table class is designed to cut storage costs while delivering the same performance, durability, and scaling. DynamoDB Standard-IA is now available in all Regions, except China and AWS GovCloud (US). For more on pricing, head here.
Amazon Inspector gets a relaunch
Amazon announced the relaunch of Amazon Inspector, which debuted back in 2015. This service can be used to automate security assessment and management to help organizations with security and compliance requirements for workloads deployed to AWS. Vulnerability management has changed a lot in the last six years, so AWS has built Amazon Inspector back up from scratch to do what people actually need in an enterprise environment without additional friction.
AWS aims to reduce cloud barriers, expand industry-specific focus
AWS CEO Adam Selipsky talks about how AWS is working to eliminate cloud barriers for users and businesses through ease-of-use and industry-specific tools.
ECR Pull Through Caching
Amazon announced pull through cache repository support in Amazon Elastic Container Registry. This solves another point of security friction in enterprises.
Many big customers have hard rules around things like using code from outside or private systems, so AWS came up with a solution for container images. Instead of needing to write scripts and manage them to keep copies of public images up to date, AWS now does it automatically.
Greengrass integrated with SSM
Last year, AWS launched AWS IoT Greengrass 2.0, an open-source edge runtime and cloud service for building, deploying, and managing device software and applications. Now, AWS has rolled out the ability to securely manage your Greengrass edge devices using AWS Systems Manager.
This makes sense. IoT is hard to manage, so AWS made Greengrass. IoT is harder to manage at scale, so they integrated it with SSM.
Now you can manage both your servers and IoT devices at scale using the same tooling. You could already do this through various custom means, but AWS has simplified and streamlined the process, so you can focus on the fun stuff rather than undifferentiated heavy lifting.
AWS Cloud WAN now in preview
AWS announced the preview release of AWS Cloud WAN, a new service that allows customers to build, manage, and monitor global wide area networks. It includes a dashboard for monitoring network health, security, and performance.
Ryan and Faye’s top re:Invent 2021 picks
OK. That was A LOT of news. But was the single biggest announcement out of re:Invent 2021?
“My top AWS re:Invent 2021 announcement has got to be all the new serverless offerings — like Amazon EMR Serverless, Amazon Redshift Serverless, and Amazon MSK Serverless (the managed streaming for Kafka,” Ryan said. “This is going to be very cool for startups in particular. We built A Cloud Guru using serverless technology, so it's always going to have a special place in my heart.”
“For me, it’s definitely SageMaker Canvas — drag-and-drop for building machine learning models — and SageMaker Studio Lab,” Faye said. “I just think it's so cool that AWS is democratizing machine learning and making it easier for everyone to get on board, and I’m looking forward to playing with both of those.”