AWS re:Invent in Review - Part 1

December 15, 2020

AWS re:Invent is one of the most anticipated events of the year for the global cloud computing community. From major product announcements to technical sessions and partner expos, the event covers both the year in review and what's yet to come for Amazon Web Services.

With our extensive work on the cloud, there’s no way we’d miss out on what’s happening! We’ll be going over all the major announcements in easy-to-follow weekly roundups, so it’s easier to digest. Keep an eye out for more extensive coverage about some of these announcements early next year!  

Week 1 AWS re:Invent Coverage

Amazon SageMaker Integrations

AWS has put in a concerted effort towards integratingMachine Learning (ML) into modern developmental lifecycles with a platform thatstreamlines the process – Amazon SageMaker. SageMaker Autopilot, a tool thatautomates the creation of ML models, hasnow been incorporated into many of its chief data management services:

  • ML for Amazon Aurora – RelationalDatabase developers can add ML capabilities to an enterprise application usinga simple query. When you run an ML query in Aurora, it can access a range of MLmodels from Amazon SageMaker and Amazon Comprehend. With a tighter integrationbetween Aurora and every AWS ML service, Amazon pegs the throughput at 100times faster than moving data between Aurora and SageMaker/Comprehend withoutthis integration.
  • ML for Amazon Redshift – Amazon showcaseda preview of the capability to run ML algorithms on Amazon Redshift data directly!There’s no need to manually select, build, or train an ML model. When you runan ML query in Amazon Redshift, all selected data is securely exported toAmazon S3 before the training data is cleaned, a model created, and the bestmodel applied by SageMaker Autopilot. All these processes are abstracted away andwill happen automatically. Once the model is trained, you can use it as an SQLfunction.
  • ML for Amazon Neptune – Graph NeuralNetworks are now available for all graph data on Amazon Neptune (using the DeepGraph Library or DGL). This library is specifically built to run deep learningon graph data. It improves the accuracy of most predictions by over 50%compared to traditional ML techniques on the same dataset.

AmazonSageMaker Data Wrangler

Data preparation has long been one of the mosttime-consuming practices when it comes to the ML process. In fact, DataScientists say that it takes up a whopping 80% of the total time they spendworking on ML problems. SageMaker Data Wrangler is a new update to AmazonSageMaker that allows you to prepare data for ML applications much faster usinga visual interface. With just a few clicks, you can connect to data sources (AmazonS3, Athena, Redshift, AWS Lake Formation), explore and visualize data, apply a built-intransformation, export the resulting code to an auto-generated script, and runit on managed infrastructure.

AmazonSageMaker Pipelines

AWS is bringing the power of DevOps to ML projects with theSageMaker Pipelines capability for Amazon SageMaker. This capability will makeit easier for data scientists and engineers to build, automate, and scale machinelearning pipelines using best-in-class DevOps practices. As we’ve come toexpect from SageMaker, all infrastructure is fully managed and doesn’t requireany work on your end.

AmazonElastic Container Service (ECS) Anywhere and Elastic Kubernetes Service (EKS)Anywhere

While AWS Fargate and AWS Outposts did offer customersmuch-needed flexibility for container deployment, some needed a bit more. Amazonhas answered the call. It understands that someone all-in on the cloud mightstill be bound by financial, legal, regulatory, practical, or eventechnological constraints that require them to deploy containers outside AWSregions. Amazon ECS’ and EKS’ simplicity will now be extended to deploy nativeECS/EKS tasks in any environment, including customer-managed infrastructure, in2021.

AWS GlueElastic Views

Glue Elastic Views facilitates building materialized viewsthat combine and replicate data across multiple data stores without any customcode. You can use regular SQL to create a virtual table from numerous differentsource data stores. Amazon announced support for Amazon DynamoDB, S3, Redshift,and Elasticsearch in the preview that’s now available across some regions. Theannouncement does promise support for Amazon Relational Database Service (RDS),Aurora, and many more soon.  

AmazonCodeGuru Feature Updates

Amazon CodeGuru is a developer tool that helps you improvecode quality using ML-powered recommendations and automated code reviews.

Amazon announced three new features for CodeGuru Reviewerand CodeGuru Profiler:

  • Python Support for CodeGuru Reviewer andProfiler – Developers working on Python applications can now use CodeGuru toimprove their code. This is a welcome update to CodeGuru – expanding itscapabilities past Java code and applications.
  • Security Detectors for CodeGuru Reviewer – Withthe addition of a new set of detectors, Amazon CodeGuru Reviewer identifiessecurity vulnerabilities and checks for security best practices in your Javacode. The security detectors can find security vulnerabilities in the top 10OWASP (Open Web Application Security Project) categories, like weak hashencryption.
  • Memory Profiling for CodeGuru Profiler – Thisupdate is focused on hunting for memory leaks and optimizing how yourapplication uses memory. It offers a new visualization of memory retention – atimeline graph that makes it easier to spot trends and peaks of memoryutilization per object type.

AQUAfor Amazon Redshift

AQUA, or Advanced Query Accelerator, is a brand-newdistributed and hardware-accelerated cache for Amazon Redshift that delivers upto 10x faster query performance. It accelerates Redshifts queries by running data-intensivetasks (like filtering and aggregation) closer to the storage layer. Amazonannounced that AQUA-powered Redshift is 100% compatible with Redshift RA3 instances,and you won’t need to change any code to make use of these performanceimprovements.  


Amazon also announced an end-to-end system that uses MachineLearning to detect abnormal behavior in industrial machinery – allowing you to implementpredictive maintenance and reducing unplanned downtime. This conditionmonitoring service will track your equipment’s health using sensors to ensure youcan easily capture relevant data.

AmazonQuickSight Q

Amazon QuickSight, a powerful Business Intelligence servicefor the cloud, also got an exciting update during AWS re:Invent – QuickSight Q.It empowers business users to ask questions about their data using everyday naturallanguage to get answers in seconds. No coding, no formats, or syntaxes tofollow. Q uses Deep Learning and Machine Learning to understand user intent andderives meaning from underlying data to generate comprehensive answers usingrelevant visualizations.

Updates are already flying in from Week 2 of AWS re:Invent. We’ll get some coverage on those announcements to you soon!


Related Blogs