The annual AWS re:Invent conference, held in Las Vegas in 2023, again showcased a remarkable array of innovations that hold the potential to shape the future of cloud computing. From next-generation compute and storage solutions to cutting-edge AI and machine learning advancements, AWS unveiled a plethora of features that are set to transform the way businesses operate and innovate.
A team of PCG consultants and engineers attended the conference and recapitulated some of the most important announcements of the week. From Nov 27th to Dec 1st, more than 125 features and services have been announced, which mainly fall into the following categories:
- Generative AI
- Machine Learning & AI
- Data & Analytics
- Compute & Databases
We have picked a few features that – we think – are relevant for our customers in moving forward with the cloud in 2024
AWS Q: A Unified Interface for Intelligent Automation
One of the most significant announcements was the launch of AWS Q, an AI-powered assistant that simplifies the process of automating tasks and processes across multiple AWS services and applications. It supports users in the interaction with the AWS console, guides them through debugging and empowers them to create and manage a comprehensive library of automated workflows. This helps in streamlining operations, reducing manual effort, and enhancing overall efficiency.
Generative AI: Empowering Creativity and Innovation
AWS also made a significant step forward into the realm of generative AI, introducing a suite of new tools and capabilities that are poised to revolutionise creative applications. These advancements include Titan Image Generator, a Text-to-Image AI, which generates photorealistic images from text descriptions, and Model Evaluation on Amazon Bedrock, which allows for a comparison of foundation models.
Machine Learning & AI: Lowering the Barriers
Easier access to machine learning services and a lower learning curve in developing ML models appear to be a focus topic for AWS. Amazon SageMaker Canvas is a visual interface allowing users to build, train, and deploy ML models without writing any code. This interface uses drag-and-drop components and interactive visualizations to make the ML development process more intuitive and accessible. With Amazon SageMaker Clarify, developers and data scientists are empowered to build responsible and trustworthy ML models by giving insights into model bias and predictions.
Last but not least, AWS has not been resting on their laurels with regards to Amazon CodeWhisperer. The service has been around for some time already, but it now offers AI-enhanced code remediation for several languages and support for IaC (TerraForm, CDK, CloudFormation) to drive efficiency in coding.
Data & Analytics
What do Aurora, Redshift, OpenSearch, and ElastiCache have in common? All of them are offered as serverless services by AWS, with the latest addition being ElastiCache Serverless for Redis and Memcached. Amazon’s Aurora service has also received a major feature upgrade: Aurora Limitless (for PostgreSQL) is now in preview state, allowing for horizontal scaling to support exorbitant numbers of write transactions per second. AWS also announced a couple of services, which now offer zero-ETL integrations with Redshift and S3, thus reducing complexity through ETL pipelines.
Compute (& Security)
Yes, we know. Compute is boring… or is it? AWS has announced the new generation of ARM-based CPUs, Graviton4, that will offer significant performance benefits compared to the current generation.
While not strictly being compute services, AWS GuardDuty and Inspector have also – again – received important improvements, now offering runtime detection in ECS, AI-assisted code remediation for Lambda functions, and scanning of EC2 instances without the need for agents (i.e., SSM agent).
Have any of you ever actually worked with AWS WorkSpaces or AWS AppStream? Even if you haven't, there is now one more reason to think about that – AWS announced the AWS WorkSpaces Thin Client, a small box that connects to an AWS WorkSpace or AppStream instance to provide the user with a Desktop experience without storing anything on the thin client itself. We were able to test the device hands-on, and keeping in mind it’s the first generation, the first impression was surprisingly good. Sure, 1080p is not state of the art anymore, but the little device seems predestined for a simple desktop in an office that is part of an environment with strict compliance standards.
Storage: It’s all about Performance and Costs
On Tuesday, Nov 28th, Adam Selipsky unveiled S3 Express One Zone, a new storage class designed to provide exceptionally low latency for frequently accessed data. This purpose-built storage solution is particularly well-suited for mission-critical applications that demand near-instantaneous data retrieval. Not only has S3 ramped up in speed, but EFS has had its feature set expanded too, reaching up to 250.000 read IOPS and 50.000 write IOPS per file system. A new storage class has also been announced for EFS (EFS Archive), that helps cut storage costs for infrequently used files that still need to be accessible over the NFS protocol.
One of our favourite storage services, FSx for ONTAP, has also received a heavily requested feature – as of Nov 26th, ONTAP Multi-AZ file systems can be set up in shared VPCs.
Last but not least, a new feature that does not really fit into any of the categories above: CloudFormation now supports deployments through git-sync. We appear to have missed that announcement at the re:Invent itself, but nevertheless, we look forward to trying this new feature. Deploying CloudFormation templates can be a real hassle, especially tracking changes. We are excited about how this feature might relieve us of some of this pain and how it behaves in hands-on.
Hopefully, the information in this post is helpful to some of our readers. Make sure to keep following us for more news.