At re:Invent, AWS Machine Learning Chief Tackles What Makes a Good Data Strategy
"Data is the genesis for modern invention," said Swami Sivasubramanian, vice president of data and machine learning at Amazon Web Services (AWS), during his re:Invent keynote in Las Vegas, Nev. on Wednesday.
Sivasubramanian's presentation focused on AWS product advancements that help organizations cultivate healthy data estates. It continued a theme started on Tuesday by AWS CEO Adam Selipsky, whose own keynote focused on the ever-expanding "data realm."
There are three characteristics of a good data strategy, according to Sivasubramanian: future-proofed, integrated and democratized. Each of the product announcements during his presentation, the highlights of which are described below, is designed to meet at least one of these requirements.
- Amazon Athena, AWS's serverless query service, now has support for Apache Spark. This news comes a day after Selipsky's announcement that Apache Spark would also be supported in Amazon Redshift.
- Now generally available is the Amazon DocumentDB Elastic Clusters managed service for auto-scaling document workloads.
- A preview is now available of a capability in Amazon SageMaker to help organizations spin up machine learning models from geospatial data.
- Amazon Redshift Multi-AZ, designed for highly business-critical analytics jobs, has reached preview.
- PostgreSQL security is getting a boost with the release of the open source Trusted Language Extensions project.
- Amazon GuardDuty RDS Protection, now in preview, will bring GuardDuty's threat-detection capabilities to Amazon Aurora.
- To help businesses automate data quality monitoring tasks like rules-creation, AWS is releasing to preview AWS Glue Data Quality.
- Centralized Access Controls, a new feature in Amazon Redshift that works in concert with AWS Lake Formation, is now in preview.
- Amazon SageMaker is getting a trio of machine learning governance features, all generally available: role manager, model cards and model dashboard.
- To advance AWS's "zero-ETL future" vision that Selipsky described Tuesday, the company has released a preview of an "auto-copy from S3" feature for Amazon Redshift.
- AWS is expanding the number of data sources its products are compatible with. An additional 22 new connectors are coming to Amazon AppFlow, and over 40 to Amazon SageMaker Data Wrangler.
- To advance machine learning education in under-privileged communities, the AWS Machine Learning University will launch classes for educators in community colleges.
AWS's various data analytics, machine learning and database services have over 1.5 million customers, Sivasubramanian said during his keynote.