New AWS CEO Selipsky unveils new AWS services at Re:Invent

While a few new services announcements were made Monday at the Global Partner Summit, the CEO keynote got the lion’s share, including ones like new Graviton3 ARM chips and related C7g Instance for EC2, AWS Mainframe Modernization to encourage mainframe migrations, and the no-code Sagemaker Canvas.

Adam Selipsky onstage at Re:Invent 2021

On Tuesday, Adam Selipsky, who replaced longtime AWS head Andy Jassy when Jassy was promoted to the big Amazon chair earlier this year, delivered his first AWS Re:Invent keynote as CEO. While his keynote, at two hours, was shorter than a typical Jassy opus, which generally checked in at a highly entertaining three hours, it was dedicated to the same thing that always made up the majority of a Jassy keynote, the inventory of major new services announcements.

Selipsky began by running through AWS’s early history, covering the impact of the AWS EC2 instance, and the AWS-designed ARM-based Graviton and Graviton2 chips developed for it.

“You still needed more however, so we are announcing Graviton3 – the next generation of AWS-designed ARM chips,” he said. The Graviton3 is 25% faster than its predecessor, and has up to 2x higher floating-point performance, and up to 2x faster cryptographic workload performance.

Selipsky then announced C7g Instance for EC2, which has DDR5 memory and 25% more compute power than Graviton2-powered instances. It is available in preview now.

“This is the first of a new federation of instances powered by Graviton3 – with more instances on the way,” Selipsky declared.

Selipsky then announced a new EC2 instance for AWS’s Trainium machine learning chip, TR1 Instances powered by Trainium.

“This is for the fastest training of machine learning models in the cloud,” he said.  “TR1 is the first training instance with up to 800 gbps bandwidth. Now with Trainum and Inferentia, customers have the best price performance for training and inference.”

From there, Selipsky moved to the announcement of AWS Mainframe Modernization, which is targeted at organizations still running mainframes who would like a relatively painless way to migrate their mainframe applications to the cloud.

“This is a service that makes it faster to modernize and run mainframe applications on AWS,” Selipsky said. “It cuts the time to migrate mainframe applications to the cloud by two-thirds.”

Next came the announcement of AWS Wavelength, a 5G edge computing service, to new markets including Canada.

“To bring a bit of AWS on-prem, with the same APIs, we built Outposts which launched two years ago,” Selipsky said. “This kind of thing has to be AWS not like AWS. We have smaller Outpost form factors which are GA today. You can do work in remote places like oil fields which may not even have connectivity with our Snow devices, and you can use Wavelength for 5G networks. Wavelength’s GA was announced last year in some places. Now we are expanding it to more, including a partnership to deliver it in Canada with Bell Canada in 2022.”

Next came the launch of AWS Private 5G, which also has a significant hardware component to it.

“This is a new service to let you set up your own private mobile network in days instead of months,” Selipsky said. It comes with all the hardware needed to build out the required network, including SIM cards, servers and small cell radio units. It also has pay-as-you-go pricing, and because it operates on a shared spectrum, has no separate per device charge.”

Selipsky also declared the GA of two new enhancements for Lake Formation, its service for building data lakes on S3.

“We are announcing the GA of row and cell level security for Lake Formation, to   restrict access to specific rows and columns,” he said. “With this, you can define access to specific rows for specific users, to put the right data in the lands of the right people, and only the right people.”

The other Lake Formation enhancement in GA is Transactions for Governed Tables in Lake Formation.

“This lets you create a governed table that supports ACID transactions,” Selipsky said. “This eliminates the need for custom error handling code or batching updates. You can keep updating data in real time while users are querying data.”

New serverless and on-demand analytics options for four analytics services – Redshift, EMR, MSK and Kinesis on Demand – were also announced.

“These new serverless and on-demand options are great for workloads where you don’t want to plan capacity,” Selipsky said. “They provide incredible performance and capabilities, while allowing you to take your analytics serverless.”

One of the most important announcements was dropped right near the end of the keynote. Selipsky announced a new capability for its Sagemaker machine learning application development tool in Sagemaker Canvas.

“This provides point and click capabilities with no coding required, allowing creation of Machine Learning predictions without any Machine Learning experience or writing any code,” Selipsky said. “This will enable a whole new group of users to leverage data and use Machine Learning to create new business insights.”

Selipsky concluded by announcing several industry-specific collaborations. Goldman Sachs Financial Cloud for Data Collaboration will reduce time and development resources required for finserv companies. AWS IoT Twinmaker  makes it easy to create digital twins of real world systems. IoT FleetWise is for automakers, and makes it easier for them to collect transform and transfer data from large fleets of vehicles in the cloud in near-real time.

“What all of these industry-specific use cases do is put the power of AWS in the hands of more users,” Selipsky concluded. “We hope that it makes it possible for many more users to become pathfinders in their companies.”