M+E Daily

Amazon Web Services (AWS) CEO Adam Selipsky used his AWS re:Invent keynote on Nov. 30, the second day of the conference in Las Vegas, to announce a broad range of new AWS products and services.

First up during the keynote, which was streamed online live also, were new AWS Graviton3 processors designed for compute-intensive workloads including high performance computing (HPC), batch processing, electronic design automation (EDA), media encoding, scientific modeling, ad serving, distributed analytics and central processing unit (CPU)-based machine learning (ML) inferencing.

Amazon announced the first generation AWS-designed Graviton processor in late 2018, and followed it up with the second generation Graviton2 a year later. AWS customers now make use of 12 different Graviton2-powered instances including the new X2gd instances (virtual environments), announced earlier this year, that are designed for memory-intensive workloads, according to the company.

Meeting the Demand for More Instances

“When we introduced” Amazon Elastic Compute Cloud (Amazon EC2) in 2006, the AWS web service that provides secure, resizable compute capacity in the cloud, “the concept of compute on demand that you could buy with a credit card and pay only for what you used… was very different,” Selipsky said in the keynote. “And it all started with an instance, so we called it the ‘instance,’ and we figured if folks needed more, they’d use more instances,” he recalled.

However, “once you started using EC2, it turned out you had something very different in mind,” he pointed out, explaining: “You wanted bigger instances with more CPU, more memory, more storage. And so then there were three” instances.

And that was “just the beginning,” he said, adding: “As more customers came onboard, we launched instances for Windows and then features like autoscaling and load balancing. But you needed more. You wanted more powerful instances to run HPC workloads, databases and analytics, and enterprise applications. So we launched entirely new families of instances. Some were compute-optimized, then storage-optimized and memory-optimized. But you needed more. So we added” graphics processing unit (GPU)-based instances and Mac-based instances.

“Compute is so foundational that customers had an almost insatiable appetite for new and specialized instances,” he noted. Therefore, AWS “continued to imagine new apps and customers were inventing entirely new businesses. This is why we now have over 475 different instance types, more than any other cloud provider, and we’re not even close to being done,” he said.

But “even as we innovated here, we realized” it was important to “rethink instances entirely [and] we needed to go deep, all the way down to the silicon” to have an impact on price and performance, he explained. Therefore, AWS “set out to do just that and started to design our own ARM-based chips,” and Graviton was born, he noted.

“Today, thousands of customers are using Graviton2-based instances and are reaping the benefits and price-performance for a very wide range of workloads, including big data analytics, game servers and high performance compute,” he said. Graviton2 offered “40 percent better price-performance than comparable x86-based instances,” he added.

But “you still needed more” and so now comes Graviton3 chips, which he said are “another big leap forward” and are “25 percent faster on average for general compute workloads than Graviton2 and they perform even better for certain specialized workloads.”

Additionally, AWS-designed Graviton3 chips offer: 2x faster floating point performance for scientific workloads, 2x faster for cryptographic workloads, 3x faster for ML workloads and use up to 60% less energy, he said.

AWS also announced three new Amazon EC2 instances powered by Graviton chips that it said help customers “significantly improve the performance, cost, and energy efficiency of their workloads running on Amazon EC2.” The new instances are: C7g instances powered by Graviton3 processors; Trn1 instances powered by AWS Trainium chips that it said “provide the best price performance and the fastest time to train most machine learning models in Amazon EC2”; and storage-optimized Im4gn/Is4gen/I4i instances based on AWS-designed AWS Nitro solid-state drives (SSDs). The latter offer the “best storage performance for I/O-intensive workloads running on Amazon EC2,” according to the company.

More Announcements

Also new, Selipsky said, is AWS Mainframe Modernization, a service that makes it faster and easier for customers to migrate mainframe and legacy workloads to the cloud, and enjoy the “superior agility, elasticity, and cost savings of AWS,” according to the company.

Using AWS Mainframe Modernization, customers can “refactor their mainframe workloads to run on AWS by transforming legacy applications into modern Java-based cloud services,” the company said in a news release. “Alternatively, customers can keep their applications as written and replatform their workloads to AWS reusing existing code with minimal changes. A runtime environment built into AWS Mainframe Modernization provides the necessary compute, memory, and storage to run both refactored and replatformed applications and automatically handles the details of capacity provisioning, security, load balancing, auto-scaling, and application health monitoring.”

Customers and systems integrators can use AWS Mainframe Modernization to help enterprise migration teams assess and analyze migration readiness and plan migration projects, the company said, adding: “There are no upfront costs for using AWS Mainframe Modernization, and customers only pay for the amount of compute provisioned.”

AWS Private 5G, meanwhile, is a new managed service that AWS said helps enterprises “set up and scale private 5G mobile networks in their facilities in days instead of months.” With only a few clicks in the AWS console, customers can
“specify where they want to build a mobile network and the network capacity needed for their devices—and AWS delivers and maintains the small cell radio units, servers, 5G core and radio access network (RAN) software, and subscriber identity modules (SIM cards) required to set up a private 5G network and connect devices,” the company said in a news release. “AWS Private 5G automates the setup and deployment of the network and scales capacity on demand to support additional devices and increased network traffic. There are no upfront fees or per-device costs with AWS Private 5G, and customers only pay for the network capacity and throughput they request.”

DISH Network, which is building the nation’s first open, secure, 5G smart network has selected AWS to enable it to “onboard and scale our 5G core network functions within the cloud,” according to Stephen Bye, DISH chief commercial officer, who called AWS a “key strategic partner in helping us deliver private enterprise networks to our customers.”

Also new is AWS IoT TwinMaker, a service that AWS said “makes it faster and easier for developers to create digital twins of real-world systems like buildings, factories, industrial equipment, and production lines.” Digital twins, AWS explained in a news release, are “virtual representations of physical systems that are regularly updated with real-world data to mimic the structure, state, and behavior of the objects they represent.” IoT TwinMaker makes it simple for developers to integrate data from multiple sources, such as equipment sensors, video cameras and business applications, and “combines that data to create a knowledge graph that models the real-world environment,” it said. “There are no up-front commitments or fees to use AWS IoT TwinMaker, and customers only pay for the AWS services used,” it added.

AWS and Goldman Sachs also jointly announced the launch of Goldman Sachs Financial Cloud for Data with Amazon Web Services, a new suite of cloud-based data and analytics solutions for financial institutions. The collaboration “redefines how clients can discover, organize, and analyze data in the cloud, thereby gaining rapid insights and driving informed investment decisions. Institutional clients will benefit from decades of Goldman Sachs experience to address data management and analytics challenges,” the companies said in a news release.