re:Invent is the biggest event on the cloud computing calendar, bringing together cloud professionals from all over the globe to hear the latest major releases from Amazon Web Services, totalling over 120 announcements over the week!
We’re here to share with you the ultimate list of the most impactful news from AWS re:Invent 2022! And to help you dig deeper into the news, we’re going to share links to some fantastic write-ups by members of the AWS Community, including AWS Heroes, AWS Community Builders, and more.
Solving Lambda Cold Starts for Java with SnapStart
Cold Starts are a bugbear of serverless computing. When a Lambda Function is first invoked, AWS will provision a runtime container to load your code, and execute the function. When the container is up and running, it can return a response very quickly! But that first invocation is usually much slower, ranging from a few hundred miliseconds, up to several seconds.
Lambda SnapStart aims to solve that problem, by shortcutting some of that initial startup process. Instead of building everything from scratch, it keeps a snapshot of the memory and disk state at the end of the `Init` phase, which can be reused. It currently supports Java using Coretto, but since it uses Firecracker’s native snapshotting capability, we may see this extended in future, and would be a huge boon to runtimes like .Net!
We’ve written up a summary of the details here on our blog at Lambda SnapStart: What we know from re:Invent 2022, and there’s a great blog detailing this release by Jeff Barr himself, titled Accelerate your Lambda Functions with Lambda SnapStart. Finally, an awesome community blog post from AWS Community Builder AJ Stuyvenberg from Datadog titled Introducing Lambda SnapShot, where he reflects on the implications for serverless developers. Both are great resources if you’re looking to know more!
Extreme Parallelization with Step Functions Distributed Map
One of the biggest paradigm shifts with Cloud Computing has been the way we handle parallelization – doing many things at the same time. Instead of taking 100 hours to do a task, we can split it into 100 pieces, run it concurrently, and have it done in one hour at nearly the same cost.
Step Functions has already had support for up to 40 concurrent executions, but with the new Distributed Map flow, Step Functions now supports up to 10,000 concurrent executions. This is so high that you’d have to boost your AWS Lambda service limits to use it all.
AWS Principal Developer Advocate Sébastien Stormacq has written a great blog post where he demonstrates using the new flow, and offers some advice on ideal use cases in Step Functions Distributed Map – A Serverless Solution for Large-Scale Parallel Data Processing.
Removing Glue Code with EventBridge Pipes
The greatest power of cloud services comes when they interact with each other to create real value. When dealing with different services, we often need to write the code that integrates the two together; which we often refer to as the “glue”. For example, we might receive messages into an SQS queue, which we want to be processed by a Step Function state machine, after we’ve filtered out the necessary messages. Up until now, we’d have to write an extra Lambda function to do that integration.
Enter EventBridge Pipes; taking it’s name from the pipe’s (`|`) we often use in scripting languages to pass information between commands. With EventBridge Pipes, we can scrap a lot of that messy glue code, configuring point-to-point integrations between the producer and consumer services of an event, including built-in capabilities for Filtering and Enriching the data!
This is one of the most interesting releases of re:Invent 2022, and the write-up by AWS Principal Developer Advocate Donnie Prakoso with New – Create Point-to-Point Integrations Between Event Producers and Consumers with Amazon EventBridge Pipes covers a full example of how it works. AWS Principal Engineer Nik Pinski, who worked on this service, also shares his thoughts as part of his Twitter thread on EventBridge Pipes.
Building Visually with AWS Application Composer
Serverless solutions can be incredibly intimidating because of the number of moving parts involved, and the way that they interact together. Least of all because when you’re building solutions across distributed teams at an agile pace, documenting all of that information in an accessible way is no small feat, let alone building the Infrastructure as Code solutions behind it.
AWS Application Composer provides a visual designer that allows you to build serverless applications through a drag-and-drop interface, which shows both the individual components, and also the links between your resources. By generating a CloudFormation SAM template, your architecture diagram actively keeps up with changes your code. If this sounds like Stackery, that’s no coincidence either!
Builds a developer experience that makes adopting serverless solutions easier is a huge plus, and the blog post by AWS Principal Specialist Solutions Architect Luca Mezzalira titled Visualize and create your serverless workloads with AWS Application Composer is a great starting point. AWS Community Builder Danielle Heberling has also done a great write-up in her blog post, AWS Application Composer, the App Building Future We Need.
Vulnerability Scanning for Lambda Functions with Amazon Inspector
Do you know all of the dependencies running in your code? If you’ve ever seen the total size of a node_modules directory, you might think twice. Last year’s Log4j vulnerability should leave us all wary of potential vulnerabilities in our software in our deep nest of dependencies. Amazon Inspector has been a handy option for years, which now includes integration with AWS Lambda Functions.
Once you’ve enabled Amazon Inspector, it can start scanning your Lambda functions for known vulnerabilities, making it easier to find and remediate the biggest issues. Amazon Inspector saw a huge uplift last year, and this is a very welcome addition. If you don’t already have a source code vulnerability management solution in place for your Lambda functions, definitely check this out, even if to find out what’s already there which you may not know about.
The official blog post by AWS Principal Developer Advocate Marcia Villalba titled Amazon Inspector Now Scans AWS Lambda Functions for Vulenerabilities gives a solid overview of the feature and it’s key points, and you may also want to check out the blog post Amazon Inspector can now scan AWS Lambda Functions by AWS Serverless Hero Jones Zachariah Noel who shares his experience going through the process!
Simplifying Development with Amazon CodeCatalyst (Preview)
When starting a new project, there’s always time spent setting stuff up. Source code repositories, folder structures, CI/CD pipelines, issue trackers, documentation, and even more. CodeCatalyst gives you the ability to create new projects, building and preparing everything you need to jump straight into writing the code. CodeCatalyst provides a number of blueprints out of the box, which are both very customizable, and allow you to have your projects built in a standard way.
This is really about removing some of the undifferentiated heavy lifting, and provides the opportunity for much greater consistency without hard work. This fills a very similar gap to AWS CodeStar which was released five years ago, but never seemed to get very far. I’m planning to try out CodeCatalyst for an upcoming project, but how it works into the future, time will tell!
AWS Senior Developer Advocate Steve Roberts covers a lot of the different features available in CodeCatalyst in the blog post Announcing Amazon CodeCatalyst, a Unified Software Development Service (Preview). AWS Community Builder Jatin Mehrotra has also explored the tool, showcasing the process of building a project in his blog post Using AWS CodeCatalyst: I Created and Deployed a React SPA #reinvent.
Flexible Provisioning with Amazon OpenSearch Serverless (Preview)
OpenSearch Serverless is now available in Public Preview, which takes away the need to manage specific clusters, and instead creating your collections as needed, and letting AWS handle the provisioning and scaling for you, which is useful if you’re dealing with large dynamic workloads.
Editorial Opinion: I’m not a fan of this service being referred to as “serverless”, since it is still very specifically priced by duration of compute units, rather than per requests or a similarly elastically fitting work unit. This service running at minimal levels currently costs a good $700+ per month, which isn’t cheap. Instead, it’s better described as autoscaling, and useful for those very large and unpredictable workloads. The service is still in preview and prone to change. Much to their credit also, members of the service team have been actively engaged in social media discussions about this, and are acknowledging this view is out there.
If you’re running this type of large dynamic workload, it’s worth checking out the blog post by AWS Principal Developer Advocate Channy Yun, author of Preview: Amazon OpenSearch Serverless – Run Search and Analytics Workloads without Managing Clusters.
Abstracting Network Complexity with Amazon VPC Lattice (Preview)
Networks underlie everything we build in modern computing, and modern cloud is no exception. Network Engineers need to be able to control and scale their networks, while developers are constantly building on them. VPC Lattice introduces a new layer of abstraction called a service network, simplifying the connections across VPCs and accounts, making it easier to build services that exist across complex network topologies.
The natural concern with abstraction is that it results in a loss of granular control. In Dave Brown’s (AWS Vice President for Amazon EC2) session, he reiterated how important it is that network administrators still retain the tools they’re used to, like Security Groups, Flow Logs, and Traffic Mirroring. While this is still only in Private Preview in a single region (Oregon / us-west-2) and it’ll be a while before we see it widely used, it holds some interesting promise.
AWS Chief Evangelist for EMEA Danilo Poccia gives a great walkthrough of building a hybrid serverful/serverless service using VPC Lattice in his post Introducing VPC Lattice – Simplifying Networking for Service-to-Service Communication (Preview). If you’re struggling with network complexity, this is one worth watching.
Analytics across boundaries with Amazon DataZone
Every modern organization relies on data. While the old paradigm was that you may have a specific team or teams that handled of all of your data across it’s lifecycle, but the truth is that data is now generated, maintained, and used across your whole organization. Don’t believe me? Ask your marketing department.
Amazon DataZone has been announced under a Private Preview as an upcoming solution to enable your data producers to share what they have, and connect them with your data consumers to access it in a way that’s relevant to your business, and within the confines of your data governance controls.
AWS Head of Product for Amazon DataZone Shikha Verma gave an awesome example of a use case within an organization and demonstration of the new tooling during the Data and Machine Learning keynote (between 1:05:38 to 1:14:35), which I definitely recommend. AWS Serverless Hero and DataEngBytes Conference Organizer Peter Hanssens described it as the biggest announcement of re:Invent 2022 for anyone in the data engineering space, so it’s definitely one to watch!
Governance Solutions for Machine Learning
It’s easy to forget just how much machine learning and algorithms drive our world. AWS gives us access to extraordinarily powerful tools, and it’s easy to miss the impacts of what we’re building. As builders and technical leaders, we need to be aware of how we’re using Machine Learning, and that we’re doing so both effectively, and responsibly.
The new ML Governance Tools for Amazon SageMaker gives us a range of tools to help govern our SageMaker solutions and their usage. We also have the new AWS AI Service Cards which gives us guidelines both on how to build for with services like Rekognition and Textract for maximum responsible effectiveness.
AWS Principal Developer Advocate Antje Barth’s blog on New ML Governance Tools for Amazon SageMaker – Simplify Access Control and Enhance Transparency Over Your ML Projects covers these new solutions, giving you a place to orient yourself on these new tools
10,000ft view with SageMaker Geospatial (Preview)
When Amazon Location Service was released last year, it felt strange that we didn’t have more integrations with the powerful suite of AWS Machine Learning capabilities. Here, AWS is opening the door to that, with Amazon SageMaker Geospatial being released to Public Preview.
Being able to process imagery with solutions like cloud removal, geomosaicing, and different statistical functions, it’s bringing together a lot of ways to augment some of that data that already exists, plus the Vector Enrichment options, like reverse geocoding. From my own experience in the Emergency Management sector as a GIS Operator, these have some really handy possibilities. It’ll be interesting to see how AWS intends to enhance our options with this service, and how it will compare to GIS Solutions like Esri’s ArcGIS.
Check out some of the capabilities now available through the AWS Blog post Preview: Use Amazon SageMaker to Build, Train, and Deploy ML Models Using Geospatial Data.
Enhancing Logistics with Amazon Supply Chain
Understanding, handling, and analyzing your supply chains is a critical function of any business. And as this decade has shown, global supply chains are incredibly fragile, and being able to identify our risks is the first step to managing them. With Amazon being the world’s largest retailer, it’s no surprise that they leverage AWS technology to support their logistical operations, and now you can too!
Amazon Supply Chain works with your existing ERP and supply chain solutions, unifying the information and providing an overlay powered by AWS’s machine learning services and insights. It’s still in preview, and will be interesting to see how this develops. Check out the information on the AWS Supply Chain service page to find out more.
Global Multiplayer Gaming with GameLift Anywhere
Multiplayer game developers have always fought with one enemy. The technical term is latency, but gamers know it best as “lag”, and the eventual screams of frustration. GameLift gave the ability to deploy and automatically scale multiplayer game servers through AWS, but developers also want to be able to deploy servers on their own hardware (imagine testing your latest patches!).
Amazon GameLift Anywhere brings the capability to run GameLift servers on your own hardware. This could potentially be leveraged to run on other cloud providers as well that may be closer to your player markets, but this would require some more custom engineering. The blog post on Introducing Amazon GameLift Anywhere – Run Your Game Servers on Your Own Infrastructure covers much of the new feature set.
Simulations at-scale with SimSpace Weaver
The real world is vastly complex, and trying to simulate it in a computer is intensely computation-heavy. SimSpace Weaver is a new solution that plugs directly into commonly used engines like Unreal Engine 5 and Unity, to scale, orchestrate, and synchronize a fleet of EC2 instances to run your simulations at an enormous scale.
This service is really targeted at a specific user base who need to simulate millions of entities at-scale, like modelling traffic flows across an entire city, simulating disasters, or the movement of millions of people. Some prominent use cases, and an example of how to get started are covered in the new blog post New AWS SimSpace Weaver – Run Large-Scale Spatial Simulations in the Cloud.
Bioinformatics with Amazon Omics
Genomics presents some unique challenges in computing, mainly because of the immense amount of data involved. Trying to store, access, and process the data in meaningful ways is something where cloud computing brings unique opportunities, although it’s still very complex. Amazon Omics gives a handle and process this information to generate valuable insights.
Simplifying the processing of petabytes of genomic data through solutions like Amazon Athena or Amazon Sagemaker opens many opportunities. If you’re looking for a solution in this space, the AWS Blog post on Introducing Amazon Omics – A Purpose-Built Service to Store, Query, and Analyze Genomic and Biological Data at Scale is a great starting point.