Featured Stories

Filter By Categories

re:Invent 2019 - new and notable announcements

By ACG Technical Editors Team  |  December 03, 2019  |   Machine Learning AWS re:Invent   |  

AWS re:Invent 2019 was packed with announcements and new features revealed. To help ease your cognitive load, we’re giving you our rundown of the most significant news from Las Vegas, so you can get a sense of what’s what, why it’s awesome, and what it could mean going forward.

1. Need for speed? New EC2 instance types get Nitro

What is it?

AWS is adding some Nitro to their EC2 instances. Amazon announced AWS Nitro System is the underlying platform for the latest generation of EC2 instances. 

Why’s this awesome?

This brings speed and power to the mix for those who need a little extra umph to their workloads. These new instances types (C5/C5d/C5n, M5/M5d/M5n/M5dn, R5/R5d/R5n/R5dn, and P3dn) enjoy the benefits of 36% higher EBS-optimized instance bandwidth all the way up to 19 Gbps. With their new High Memory counterparts now also supporting a 36% increase to 19 Gbps of EBS-optimized instance bandwidth.

What’s next?

What do all those wonderful numbers mean? 

  • You can speed up sections of your workflows based on the EBS-optimized instance’s performance.
  • You can have smaller instances that still meet your EBS-optimized instance requirements — a real money saver.
  • You can also handle unplanned spikes in EBS-optimized demand without your application getting sluggish as we were feeling on the fourth early morning of re:Invent before coffee.
2. Amazon opens The Citadel for knowledge-hungry gurus

What is it?

For would-be AWS Maesters out there, Amazon has unveiled the The Citadel of AWS knowledge. The Amazon Builders’ Library is a repository of living in-depth articles from senior technical leaders. These articles detail how Amazon plans and operates the software and infrastructure behind a couple of properties you may have heard of before, Amazon.com and AWS. 

Why’s this awesome?

If you’re a cloud guru who wants to learn from and better understand how Amazon operates, this is your chance. The Amazon Builders’ Library has articles on topics like overcoming challenges around distributed systems, carrying out health checks, and implementing workload isolation using shuffle-sharding.

What’s next?

At the moment, there are only 13 articles, so it isn’t a massive reference library yet. But Amazon says they plan to update it regularly. This has great potential, both as a learning tool to understand AWS architectures and as a guide of best practices. 

3. CodeGuru gives your code an extra set of automated eyes

What is it?

Amazon CodeGuru (hmm … something about that name has a nice ring to it) is a machine learning service to automate code reviews. It provides recommendations around performance improvements and best practices. Just add Amazon CodeGuru to your GitHub or CodeCommit Pull Requests, and it’ll pour over the code. The review process is based on millions of code reviews and thousands of applications profiled on open source projects and internally at Amazon.

Why’s this awesome?

Some teams may choose to adopt this as an essential part of their workflows. Others may use it as a tool for benchmarking their own code reviews. If it works as Amazon suggests it might, this could become a very powerful tool for development projects, large and small.

What’s next?

It currently only supports Java, but more languages are planned. If Amazon can provide ways to create your own lint rules, it could become even more powerful.

4. Cross the streams! Kinesis Video Streams now two-way

What is it?

Kinesis Video Streams now supports two-way communication via WebRTC. This enables applications such as video chat and peer-to-peer streaming.

Why’s this awesome?

It’s built on WebRTC, an open source project enabling real-time communications between web browsers, mobile apps, and devices via simplified APIs. Pricing is based on the amount of data ingested — you pay only for what you use. 

What’s the use case for something like this? Think security camera monitoring from a mobile device with the ability to send audio broadcasts back to the security camera. Given the potential security issues with many off-the-shelf cameras, this may enable more organizations to maintain more fine-grained control of their infrastructure.

What’s next?

This type of infrastructure can be costly and complicated to establish. It’s exciting to see how it might be expanded, particularly through integration with the new Amplify frameworks for iOS and Android.

5. Outposts brings (some) services to the comfort of your server room

What is it?

One of the first announcements Tuesday was the general availability of AWS Outposts, which gives access to some AWS services in a local environment still managed by good ol’ AWS.

Why’s this awesome?

Outposts provide an on-prem installation for clients to have access to a limited (but still pretty cool) collection of AWS services on site. This service allows for fast localized testing of deployments and more. Word of its release was followed by a parade of feature announcements for the long-awaited service: RDS (in preview), ECS, EKS, and App Mesh.

What’s next?

With RDS announced to be in preview, this powerful localized service will eventually allow end-users to really develop, test, and deploy to real-world environments with total confidence. 

6. AWS gets local in L.A. with Local Zones

What is it?

Speaking of bringing services closer to users, AWS Local Zones is a new type of deployment that brings select AWS services closer to places with no current AWS Region. The first stop? Los Angeles. 

Why’s this awesome?

This brand new type of infrastructure deployment is a limited service extension of the AWS Region it’s associated with. It’s aimed at more high throughput and performance clients that require low latency with the scalability of cloud services. Organizations involved in media and entertainment, real-time multiplayer gaming, and machine learning should take note.

What’s next?

This opens the window for some clients who just wouldn’t have found cloud computing viable for their needs. Keep your eyes peeled. It’ll be exciting to see where this goes. 

7. Wavelength waves off 5G device-app latency 

What is it?

Hold onto your hats, 5G ultra-low latency application support is about to hit with AWS Wavelength.

Why’s this awesome?

With the introduction of 5G networks, AWS is teaming up with telecommunication providers to introduce AWS Wavelength. This embeds AWS compute and storage devices at the edge of telecommunications providers 5G networks, minimizing network hops and latency to go from your 5G device to your application. 

What’s next?

Blistering speeds are currently slated for countries around the globe but will be limited to North America, Europe, Japan, and South Korea in 2020.  

8. Transit Gateway helps VPCs and on-prem networks get cozy

What is it?

Announced Tuesday for general availability, AWS Transit Gateway network manager provides clients with a centralized gateway to manage their on-prem network and VPCs in one space.

Why’s this awesome?

Ever wished you could manage and connect your local on-premises network with your VPC? (Neither have we. But some do.) Enter AWS Transit Gateway, which allows you to do just that with ease. 

What’s next?

Also allowing for inter-region peering and IP multicasting, Transit Gateway is set to make managing and connecting your VPCs and on-prem networks a hell of a lot easier.

9. AWS unveils a slew of machine learning features

What is it?

Over the course of re:Invent 2019 day three, Amazon announced a slew of promising new features and services inspired and powered by machine learning (ML). These new additions can help with security issues, fraud detection, review of ML predictions, and AWS Compute optimization. They join existing services, such as GuardDuty and CloudWatch Anomaly Detection.

Why’s this awesome?

A promising new list of machine learning features and services? Yes, please!

  • Amazon Detective
    Amazon Detective uses machine learning to ID potential security issues in your account. It complements GuardDuty, which detects initial unusual behavior from VPC Flow logs and CloudTrail. This allows you to perform a more in-depth investigation into that behavior based on machine learning.
  • Augmented AI
    Refining your machine learning models often requires human involvement to validate its accuracy. The Augmented AI service will assist you to perform human review of the outputs of your models which otherwise may be too much for an ordinary team. Augmented AI allows for large-scale quality-checking, such as validating the text outputs of Amazon Textract, reviewing object classification, Amazon Rekognition object classification, or checking the outputs of your own models. Your first 500 human reviews from Augmented AI are within free tier, making it easy for customers to make a start refining their ML models with human reviews. 
  • Amazon Fraud Detector
    Amazon has over 20 years of experience with global online commerce, so it’s safe to assume they know a thing or two about fraud detection. With the new managed service Amazon Fraud Detector, Amazon shares its own machine learning fraud detection techniques. Customers can create a fraud detection model within minutes — no machine learning expertise required.
  • AWS Compute Optimizer
    Using AWS Compute Optimizer, customers can obtain free insights and recommendations on their EC2 instance size and autoscaling group choices. This builds upon the existing utilization checks AWS has provided for years as part of its Trusted Advisor service. Trusted Advisor appears to have gone unloved for quite some time, and Compute Optimizer continues the AWS trend of decentralizing operational advice from Trusted Advisor and moving it into separate distinct services.  

What’s next?

All the machine learning announcements made at re:Invent seem to signal a mega year for machine learning in 2020.

10. Machine learning a thing or two? AWS wants to help

What is it?

Those wanting to apply machine learning to solve their own business problems weren’t left behind. AWS announced a large number of additions to Amazon SageMaker designed to make it easier and quicker to engage in machine learning.

Why’s this awesome?

Amazon SageMaker Studio was introduced as an integrated development environment for machine learning on AWS, bringing together a number of new and old tools into a streamlined visual interface. A number of announcements were made fleshing out the feature set of this service.

  • SageMaker Experiments helps you track and compare your ML experiments. We're not quite sure why it needed a separate name (it feels like an obvious feature for SageMaker Studio) but it will certainly be appreciated. 
  • SageMaker Notebooks is in preview, providing what seems like an AWS wrapper around Jupyter Notebooks.
  • In an announcement worthy of a “Yo Dawg” meme, SageMaker Model Monitor is sort of like machine learning for your machine learning models. It monitors deviations from your training models over time and alerts you to make changes.
  • SageMaker Autopilot moved into general availability. This allows you to automatically train and improve models based on your own tabular data. This functionality has been in Azure Machine Learning Studio for a bit, but it's good to see AWS continuing to invest in this space.
  • SageMaker Debugger is designed to provide insight into what’s happening with your model training. SageMaker Studio supports both Autopilot and Debugger.

Those not using SageMaker also saw some improvements with updates to AWS Deep Learning Containers and the availability of Deep Learning AMIs. AWS Deep Java Library was also announced, bringing machine learning tools to Java. This is good for those in Java environments who can't or don't want to use Python. Given the extensive ML framework ecosystem in Python, this may take a while to gain momentum. But it's great that AWS is using its resources to support additional languages.

What’s next?

AWS making it easier for people to get some of that sweet machine learning action should be a welcome development to many. We’ll see how this plays out in 2020.

11. AWS takes a quantum leap with Amazon Braket

What is it?

Amazon has announced their new quantum computing service, called Amazon Braket.

Why’s this awesome?

Easy — it’s quantum computing! Braket offers quantum compute power through a fully managed service. If your quantum knowledge doesn’t extend far beyond Schrodinger’s cat or the traveling salesman problem, never fear! Amazon is bundling tools into Braket to help lower the learning curve, such as a simulation environment running on traditional computers, where you can build and test quantum algorithms before deploying them on one of the quantum machines.

For businesses, the Quantum Solutions Lab consulting unit offers collaborative workshops and educational resources to help engineers and business leaders get “quantum ready” and start identifying promising quantum computing applications.

What’s next?

Scheduled to be released to general availability next month, Braket will be one very interesting development to watch. At present, none of the quantum systems offered by Braket (or any other existing quantum computers) can support highly sophisticated enterprise-grade applications, so we’re probably still a ways from broad adoption. Still, it will help researchers, developers, and businesses as they start getting quantum ready.

Additionally, we’re curious to see how Braket’s “hybrid quantum/classical “ processing progresses. It may be a pitch to lower the cost of quantum computing, and bring it more into a range where corporate customers start exploring the power and potential of quantum computing for their businesses.

12. DeepRacer, meet DeepComposer! AWS launches a new Machine Learning Community Initiative

What is it?

DeepRacer has seen a host of updates this year, with new features supporting racing multiple cars on a single track, plus new racing formats.

On top of this, AWS has announced DeepComposer, a new applied Machine Learning service that lets users play a short melody, from which DeepComposer will generate a fully original song.

Why’s this awesome?

AWS is deepening its focus in these community-based Machine Learning projects, bringing more people into the space. The latest DeepRacer cars, equipped with LIDAR and stereo cameras, are bringing new challenges into the space, and DeepComposer is introducing a new type of tech, Generative AI, into an easy-to-use program.

What’s next?

DeepRacer has proven a cult classic since its introduction last year at re:Invent 2018, attracting enthusiasts across the world. DeepComposer may draw a similar fanbase. Whether users can take what they learn through these DeepX services and translate them into actionable outcomes using AI and Machine Learning remains something of an open question, however.

13. AWS makes massive sustainability commitments

After a big year for the climate change movement worldwide, sustainability is a topic on many organisations’ minds. re:Invent’s massive carbon footprint aside, AWS’ Peter Desantis today made some ambitious commitments to provide a more sustainably-minded cloud computing platform.

Desantis wrapped up tonight’s keynote by furthering Amazon’s commitment to climate action. AWS has set out to reach an 80% renewable energy goal by 2024 and plans to operate fully on 100% renewable energy by 2030.

Although traditionally lagging behind the eco-friendly street cred of both Microsoft Azure and Google’s Cloud Platform, AWS has recently been making headway. Earlier this year AWS announced three new wind farms - one in Ireland, one in Sweden, and one in the United States.

One surprising addition which brought a smile to A Cloud Guru’s Melbourne office was news of several more renewable energy projects – including the first outside of the US and Europe, right here in Australia!

14. Easier Golden AMIs and better access analysis, oh my

For those who live and breathe AWS on a daily basis, re:Invent brought some useful announcements, including EC2 Image Builder, IAM Access Analyzer, and S3 Access Analyzer.

If you create Golden AMIs, you know it can be painful. Support from AWS Systems Manager is less than ideal. EC2 Image Builder helps cut out the extra work to build a pipeline, leaving few excuses to not adopt an immutable approach for EC2 workloads. We particularly like the built-in automated testing and compatibility features with Resource Access Manager sharing. This will be helpful for both larger and smaller organizations.

Also announced: IAM Access Analyzer and S3 Access Analyzer, providing free zero-config insight into the state of your IAM policies and your S3 access policies respectively. While more security tooling is welcome, there’s already significant service sprawl, which can make it difficult to understand where to look to assess the state of your cloud security. Looking ahead, we hope to see some consolidation across the services to create a more unified experience for AWS newcomers, or those who don’t pay constant attention to AWS service announcements.

What happens in Vegas shapes the year ahead 

If re:Invent was any indication, 2020 should be an interesting year for AWS gurus. Keep tabs on it all with our original series AWS This Week. And level up your Alexa game with skill-making smarts courtesy of our new AWS Certified Alexa Skill Builder – Specialty course. Keep being awesome, cloud gurus!

Popular Tags

Enterprise Certification Career News New Products AWS Press Releases Partner Story Corporate Interviews Cloud Security Google Cloud Platform Machine Learning AWS Summit Serverless azure re:Invent Containers GCP tradeshows

Search the ACG blog