The 5 biggest re:Invent announcements to transform your business

December 19, 2019

Author: Steve McCormick

AWS re:invent 2019

Image source: AWS Events

Re:Invent is famous for announcements - big announcements. And with 77 announcements coming out of this year’s AWS conference in Las Vegas, re:Invent lived up to its reputation and delighted the 65,000 plus attendees.

CEO of AWS, Andy Jassy, kicked off the annual event on 2 December with a keynote address where he showcased the most substantial of these announcements.

In his keynote later that week, AWS’s Chief Technical Officer Werner Vogel, showed how Amazon and AWS is focused on sharing learnings and experience in building for scale and resilience.

Of all the announcements, there are five that stand out for me. Amazon has redefined hybrid computing with Outposts and Wavelength, democratised machine learning (ML) even further with SageMaker Studio and opened up more of its own experiences with Amazon CodeGuru and Contact Lens for Amazon Connect.

1. AWS Outposts

Sixteen of the 77 announcements were in the area of compute.

AWS Outposts were first talked about in 2018 and in the past year it’s clear the AWS service team have been busy. AWS Outposts is a fully managed service that extends AWS infrastructure, AWS services, APIs, and tools to virtually any data centre, co-location space, or on-premises facility for a truly consistent hybrid experience - with availability in almost all areas of the world. And yes they are in Sydney!

Outposts are available from the console, installed by AWS technical teams and managed and maintained by the AWS team. They bring local processing and ultra-low latency in a single vendor solution. The same APIs, same tools mean solutions that can span both local and cloud capacity.

AWS Outposts are not AWS in a box though; the services are limited to EC2, EBS, VPC, EKS and EMR, with RDS in preview form. The service selection is aligned to the problem outposts were designed to solve - local compute for low latency solutions. Each outpost is connected to the nearest region for access to other AWS services. For Australian customers, the great news is that we don’t need to wait - outposts are available now.

2. AWS Local Zones

We’ll have to wait for AWS Local Zones with the first launched in LA. AWS Local Zones place AWS compute storage, database and other services closer to large population centres. I’m not sure if Brisbane counts as a large population on a global scale but the use case for Australia is getting that AWS capability out of the Sydney region. Again, just like AWS Outposts, AWS Local Zones are great for latency-sensitive applications and come with more of the elastic services that we are used to from AWS.

3. AWS Wavelength

The final announcement Jassy made was AWS Wavelength. This service offering brings compute, storage, container and Kubernetes services right to the edge. It allows the deployment of applications into a wavelength zone to provide for single-digit millisecond latency.

Configured as an extension of a VPC, this capability brings the speed needed for complex IoT and the almost real-time functionality of VR and AR while this may seem like just some tech hype it has the potential to open up a new age in applications and services.

Only a few years ago, the message from AWS was to put everything in the cloud, which didn’t always work. With the launch of these three services, AWS have solved the problem - removing the latency by putting the cloud everywhere.

4. AWS Sagemaker

Data is the new oil. It’s valuable, but if unrefined it cannot really be used. It has to be changed into gas, plastic, chemicals, etc. to create a valuable entity that drives profitable activity; so must data be broken down, analysed for it to have value.” Clive Humby.

AWS also kept up the innovation in the data science and ML space. It’s opened up data science to even more people with the launch of a suite of capabilities in the Amazon SageMaker family.

The announcements in the Amazon SageMaker family are exciting and together create a powerful capability to mine pools of data for that insight. Holding all of the services together is Amazon SageMaker Studio, the first integrated development environment for ML.

This interface provides a single unified interface with all the tools to take a ML model from experimentation to production quickly. Sounds great but in the back of my mind, I still thought that all this meant was life got better for the data scientist!

As Andy Jassy began to open the box on Amazon SageMaker Studio, it became clear that AWS has created a no-code option for ML development. Amazon SageMaker Autopilot in a few clicks can turn anyone into a data scientist - well at least have a real chance of getting results that could make a difference in classification and regression ML models.

For most people, ML is out of reach. You need skills in pre-processing, model selection, model debugging and tuning and then once in production, monitoring. What we are talking about here is being able to find an algorithm that can extract patterns from an existing data set and then use that to create a predictive algorithm to help build amazing applications. This is hard work. The general answer is to find a data scientist and get them doing lots of hard science and experimentation. For non-experts, getting any results, let alone good results is really tricky.

Amazon SageMaker Autopilot is simple. Upload your dataset into a S3 bucket and point Amazon SageMaker Autopilot at that bucket. Amazon SageMaker Autopilot inspects the data and runs options to work out the optimal combination of data pre-processing steps, ML, algorithms and hyperparameters.

The brilliant capability is that this all happens with full visibility – no black boxes here. You get the python code used for pre-processing that you can tune and the notebooks covering the data exploration, and the details on the model used.

The other tools in the Amazon SageMaker Studio also provide support in taking these predictive algorithms to production. Amazon SageMaker debugger helps you troubleshoot your models as they train, optimise training times, and improve model quality. Once the model is deployed, monitoring, visualisation and drift analysis are detected via Amazon SageMaker Model Monitor, enabling you to track and improve the quality of your predictions continuously.

5. Amazon CodeGuru

Amazon CodeGuru is another service that demonstrates how Amazon and AWS are sharing their internal learnings with customers. Last year the focus was on Amazon Forecast a time-series data prediction service that Amazon uses themselves to manage their supply chain. This year it’s all about Amazon CodeGuru a ML service for automated code reviews, powered by ML trained on best practices, open-source project and Amazon’s own code reviews. This constant focus on bringing the experience of such a large global organisation into the hands of so many businesses is great to see.

As one of the largest gatherings of cloud service providers and practitioners, Re:Invent creates a concentrated education, learning and growth environment for people and businesses. It’s a great opportunity to talk and understand what our colleagues from all over the world are doing, what they are seeing, and how everyone is helping customers.

If you’re interested in discussing any of these new announcements and how they could help to solve some of your business challenges, contact Arq Group.