Reinvent 2023 just wrapped up and there have been a bunch of exciting updates that you need to know about. In this article, we’ll be recapping the top 7 announcements of the event.

S3 Express OneZone

Get ready for faster S3 performance with a new storage class called Express One Zone. This new classification offers single digit latencies and is ideal for frequently accessed and high performance applications. It’s cheaper too, coming out at 50% less than standard and can scale up to support millions of request per minute.

To optimize even more, you can co-locate your compute infrastructure in the same availability zone to minimize data transfer and optimize latency. The one drawback of this mode is that data is only located in one Availability Zone, meaning it isn’t as durable as the other classifications. This new classification is ideal for use cases like machine learning, analytics, and high performance computing.  

Aurora Limitless

RDS users will be happy to learn that Aurora now supports a LIMITLESS feature allowing infinite throughput for write operations. Previously, this feature was limited to reads, where aurora would automatically divert read heavy traffic to dedicated read instances.

With Aurora Limitless, you now have access to the same functionality but for write operations. Behind the scenes, Aurora Limitless shards your data into smaller subsets and places them on independent nodes. A Transaction router keeps track of where your data is located and performs write operations on the corresponding node. All of this is completely hidden form the end user, and allows you to focus on building your application instead of having to maintain complex solutions for scaling your data.

Aurora Limitless is available in Preview mode today.

Elasticache Serverless

For those of you looking that have read heavy applications requiring a cache, you’ll be happy to hear Elasticache is now offered in Serverless mode. This new configuration eliminates the need to set up dedicated machines you’re your cache cluster, and instead let’s AWS handle all the resource management for you.

ElastiCache Serverless uses a simple pay for what you use model where you’re only charged for the amount of data you store, and Elasticache Processing Units, or ECPUs consumed by your application.

Workspace Thin Client

Next up is a hardware announcement with the  Amazon WorkSpaces Thin Client – a new way to securely connect to remote virtual environments at a low price.

Today, many enterprise clients log on to virtual desktops with expensive laptops. Many companies are looking to reduce their cost while still allowing access to cloud resources.

The Workspaces thin client is a low cost alternative that allows you to connect to remote development environments like Worspaces, Workspaces Web, or Appstream. The product is a tiny computer, similar looking to a fire cube, and costs just $195. You can connect peripherals including your mouse, keyboard, and headset, and the device even supports dual monitor setup.

Overall, WorkSpaces thin client is low cost option to give users quick and secure access to the AWS cloud.

Step Functions External Endpoints

Next up is Step Functions support for calling external dependencies via HTTPs endpoints.

This new feature makes it easy for your step functions to directly integrate with external APIs like Strip and Github. Prior to this feature being released, needing to call an external dependency meant creating a tiny Lambda function to perform the job. Now, the same job can be done with a just a couple mouse clicks.  

This new feature simplifies dependency integration and makes it a whole lot easier to integrate your Step Functions with outside applications. And it features support for API credential management by storing your secrets in Amazon Secrets Manager.

ETL Improvements

Often applications that store transactional data for business use cases need to expose that same data for analytics and business intelligence purposes. Moving data around, or ETLing it to other destinations, can often be a time consuming process that involves setting up multiple pieces of AWS infrastructure.

This year, AWS announced mutiple Zero and near zero ETL integrations to make moving your data much easier.

First, you can now send data stored in your Aurora, RDS MySQL, and DynamoDB database directly into Amazon Redshift. Data propagation is database dependent, but as an example for Aurora, a write in your table becomes available in redshift in about 15 secodns on average. All of these options use a simple point and click setup in the AWS console.

Those of you using Opensearch will be really excited these next two. The first is a new Zero-ETL feature allows you use Opensearch to directly query data that is stored in your Amazon S3 buckets. No more setting up complex data pipelines to get the data into your opensearch cluster. Once configured, you can explore your S3 data using the popular Kibana or Opensearch dashboard to derive application insights.

Finally, you can now use Zero-ETL to load data from DynamoDB directly into opensearch. This one works slightly different than the S3 integration, since the data is getting replicated and duplicated into your cluster. Previously this used to require a ton of manual effort with DynamoDB streams, Lambdas, and other components. Its now greatly simplified and just a couple of clicks away.

Amazon Q

By far the biggest announcement of the event has been Amazon Q: a generative AI powered assistant that can be tailored to your specific business. Q is a clear competitor to popular products like ChatGPT, allowing users to get relevant answers to pressing questions, solve problems, and generate content. The Chatbot can be customized to connect to your company information. You can use it to ask questions, summarize content, and take actions based on internal documentation on company documentation or wikis.

Q is also integrated directly in the AWS console. You can use it to get advice on AWS design patterns, best practices, and other AWS specific insights. Q is trained on the AWS documentation including the well architected framework and can give you useful suggestions when in a pinch.

The most impressive part of Q is its coding assistant capabilities. It can integrate in IDEs like Visual Studio code or Jetbrains to provide development assistant like code summaries, code generation, and optimization tips. You can use it to ask questions about your current file, such as describing what a class does, to get high level understanding of a new code base quickly.

Q was talked about a LOT and a big feature of re:invent 2023. Expect more active development on this area in the coming years.