Another re:Invent is upon us, and as is tradition, Amazon has kicked things off with their Midnight Madness. For the uninitiated: Midnight Madness is a party that starts the Sunday night before the first official day of the conference to celebrate the very first slew of announcements that AWS will be making, and last night was no different. I may be getting a little too old to have stuck around for the party part of things, but here are the announcements that particularly caught my eye.
Similar to previously released AWS DeepRacer, AWS DeepComposer is a hands-on experience with machine learning in ways that many people may not think of it. Signing up for this newly released service will get you access to a specially programmed electric keyboard that connects to an AWS managed Generative Adversarial Network (or GAN) to take musical samples you input and generate new, algorithmically created tunes. This is more of an exercise, but we’re getting to see some real-life use cases for machine learning in play here to open up more potential for cloud computing. Even in its simple form, this service could be beneficial for content creators who need original music without wading through licensing or picking from a set range of audio tracks. Read more about this announcement here.
AWS End-of-Support Migration Program for Windows Server
This particular announcement is something near and dear to my heart: dealing with legacy Windows applications while migrating to the cloud. Last night AWS announced this new program, with a bundled “capturing” application that watches low-level system calls and registry key changes made when installing and configuring an application in a legacy version of Windows. The capturing program stores that information in a directory on the local machine, and then allows for the application to be migrated to a newer version of Windows (even when Microsoft’s native Compatibility Mode is unavailable). This is a migration program that requires assistance from a trusted AWS Partner like Onica, but the implications of this mean that usually complex or costly workarounds to running legacy applications in the cloud could be greatly simplified. Read the full announcement about the EMP Program on their blog post.
EC2 Image Builder
As someone who’s had to write his fair share of OS configuration scripts for automated pipelines, I found the announcement of AWS’s EC2 Image Builder very intriguing. Rather than create CI/CD pipelines for AMI creation from scratch, AWS is now offering a hosted service to create custom-configured EC2 images for Windows Server and Amazon Linux 2. This pipeline service can help ensure automatic provisioning as the images are built and become available, as well as enforce automated testing and image validation. If that sounds too good to be true, the service also hooks into EC2 VM Import/Export service to create images in formats that you can even run on-premises, eliminating time-consuming build pipelines that required multiple build stages for multiple hosting environments. Read about the service with a simple walkthrough on the AWS blog here!
Amazon SageMaker Operators for Kubernetes
Kubernetes is undeniably the container orchestration tool of choice that has taken the technology industry by storm. And containers are an extremely useful tool for many workload types; but there’s been a gap between those heavy-lifting ML workloads running in Kubernetes and the simplified toolsets available from AWS like SageMaker. With this announcement, Amazon helps bridge the gap between the flexibility of running ML workloads in k8s or Amazon EKS and the benefits of Amazon SageMaker, helping simplify the infrastructure management and take advantage of Managed Spot Training to reduce costs. This will absolutely be a boon for any shop making use of Amazon SageMaker but struggling to blend those workloads into Kubernetes workflows. Read the docs about this new Kubernetes Operator for more information.
Amazon EventBridge Schema Registry Preview
Amazon EventBridge is a service that allows for the development of event-driven software architectures by decoupling components, and leveraging a serverless event bus. This is critical when developing scalable serverless applications, but requires quite a bit of know-how to leverage. Amazon’s announcement of the Schema Registry now makes it simpler to store event structures in a central location, find existing schemas, create or import new schemas, and map them to a variety of code languages for reference. There is also a discovery feature which can add all schema sent to a bus in the registry for easier creation and importing. This service is in preview with limited region accessibility, so read the details on the AWS “What’s New” announcement page.
AWS made some other impressive announcements like a simplified BYOL interface and management experience to help eliminate some of the headaches associated with Microsoft licence compliance in the cloud, an automated license discovery tool called AWS License Manager, Amazon Transcribe Medical for healthcare-industry-specific transcription services to simplify healthcare workers’ data entry workflows, and new features for their AWS DeepRacer like object avoidance using multiple sensor inputs.
Remember, these announcements are just the tip of the iceberg that is AWS re:Invent 2019, so stay tuned for more coming out over the next week!