AWS Re:Invent 2023 Recap

CEO keynote highlights

from Dennis Faucher

CEO Adam Selipsky delivered this year's keynote, which was all about the generative AI stack. Amazon Bedrock and Amazon Q received much of the focus:

• Amazon Bedrock: A tool (and managed service) that allows users to build and scale GenAI apps and foundational models with access to emerging models from the largest AI innovators — all via a single API.
• Amazon Q: A GenAI-powered assistant (still in Preview) that will chat with workers to answer questions, solve problems, generate new content and take various actions with secure connectivity to your company's underlying data repositories, code and enterprise systems.

Combined, these tools should help developers and business users accelerate their ability to derive business results from GenAI solutions.

Of course there were many other technology-related announcements, too, including related to AWS Graviton4 and Tranium2 processors; AWS S3 Express One Zone; AWS ElastCache Serverless; AWS Code Whisperer; AWS Inferentia2; AWS Sagemaker and many more!

from Todd Barron

I was very excited to hear about the new AWS Services announcements and was not disappointed! AWS is taking great leaps forward with the expansion of AWS AI, their training capabilities and their ability to scale large language models.   

Several of the announcements should be very useful to WWT customers: 

• AWS S3 has been around since AWS has, but their new S3 Express One Zone for low latency and high-performance storage was very interesting. Lower cost and higher performance are a powerful pair in any technological advancement. We met with AWS S3 leaders and discussed several opportunities that will generate a lot of value for our customers in 2024.
• Redshift has been a great option in data analytics solutions and the Amazon Redshift Serverless edition announced will give customers greater control of their spend. Cost management is always top of mind and once this is out of Preview, I look forward to using it.
• The new Vector Engine for Amazon OpenSearch Serverless will make it easier to build ML-augmented search and generative AI applications.
• The Graviton family has been improved with the release of Graviton4 processors, powered by the new R8g instances. Graviton4 offers up to 30% better performance and has larger instance sizes with up to 3x more vCPUs than the previous generation. These instances will give our customers the power they need but at a lower cost.
• Amazon Aurora now can scale to be unlimited in size!  You can scale Amazon Aurora clusters to support millions of write transactions per second and manage petabytes of data. All of this is available without expensive licensing, and it works as a single database to calling applications. This should prove useful in several scenarios and will be very exciting to see once it's available outside of Limited Preview.
• Amazon Q, in Preview, is Amazon's own generative AI assistant and will be utilized in many areas across AWS. It is built to be secure and private, can learn from your company's information, and personalizes its responses based on your role and permissions. Look for this to show up in many services across all of AWS in the future.
• If you are leaning heavily into foundation model training, Amazon SageMaker HyperPod is for you! It enables you to train FMs for weeks and months without disruption. You can split models and training sets across AWS cluster instances for efficient scaling. It also enables a more resilient training environment with automated handling of errors and faults. Several of our FM-heavy-use customers will really benefit from this new solution.

Intelligent edge and hybrid cloud highlights

from Jenny Adefala

AWS has been leading the way in providing customers with comprehensive and flexible solutions for hybrid cloud and edge computing, with services such as AWS Outposts, AWS Wavelength, the AWS Snow Family and more. These services address various customer needs, such as low latency, data residency, migration and modernization, and AWS at the far edge. AWS customers are using these services to innovate in various industries and domains, including gaming and media, telecommunications, manufacturing and healthcare.

There was a lot of talk about generative AI, a leading driver for hybrid cloud edge solutions, and the need for low latency and high bandwidth responses. GenAI solutions can help customers unlock new possibilities, enhance user experiences, and solve complex problems at the edge. 

A few AWS re:Invent announcements that caught my attention included:

• AWS launched two new chips for AI training and inference: AWS Graviton4 and AWS Trainium2. These chips are designed to make running GenAI and other workloads faster, less expensive and more energy efficient.
• Amazon Titan Image Generator was announced to enable content creators to generate realistic, studio-quality images using natural language prompts. This can be accessed in Amazon Bedrock, a fully managed service that helps you easily build and scale GenAI applications.

At our booth, WWT showcased three solutions that leverage hybrid cloud-edge solutions in partnership with AWS:

1. Megh Computing delivers a video analytics solution (VAS) that provides a real-time, AI-based streaming analytics platform for creating actionable insights from data at the edge, using hardware architecture from edge to cloud. Examples of VAS retail can provide insights on customer behavior, store layout, inventory management and more.
2. WaitTime uses artificial intelligence to monitor crowd behavior in large-scale venues, such as stadiums, theme parks and airports. WaitTime provides real-time data on wait times, crowd densities and traffic patterns, which can help improve guest experiences, operational efficiency and safety.
3. Another Reality Studio specializes in creating immersive and interactive extended reality (XR) applications for various business industries using Unity and Unreal Engine. Using Unreal Engine, we showcased a digital twin of WWT's global headquarters in the form of a realistic 3D model that was explorable using AR/VR devices.

from Todd Barron

We had great conversations with AWS IoT leaders and have several upcoming collaborations that we are very excited to announce in 2024. Our edge solution, when combined with AWS components such as IoT Greengrass, will enable customers to have the data governance and latency required at the edge while also being able to utilize the power of AWS AI Services. AWS Bedrock, AWS Trainium2 and the fruits of the collaboration between NVIDIA and AWS will all be accessible via our edge platform and open many opportunities for customer solutions. 

Cloud cost optimization insights

from Adam Fisher

Most tech conferences usually focus on the most recent answers to "what?" and "how?" Yet many customers are also asking "why?" 

The keynote with Dr. Werner Vogels took us through a brief history of how cloud computing came about and the path enterprise customers are taking to understand how best to use it. Learning the technical nuances required for running production workloads in the cloud is a complex task. Cloud agility allows for architectures that account for things like security, availability and resiliency in compelling ways.  

These new architectures can be both effective and expensive for customers. One tenant of cloud that often gets overlooked is architecting with cost in mind. This can be hard to do when trying to bring a product to market or running mission-critical workloads. But cost optimization should be an iterative part of every enterprise cloud organization. Not only does it result in better ROI, but it helps to drive the sustainability of cloud computing.

Cost Optimization Hub is a new product announcement that brings an enhanced view of AWS cost optimization to customers. It gathers recommendations from across the existing billing and cost tools currently within AWS to give an organizational look into cost savings. This will help enable customers to iteratively improve their costs as they continue to learn how to integrate cost optimization into the design and architectural process.

Automation is essential 

from Zach Splaingard

In the sessions I attended, which spanned many technologies across a variety of industries, there was one obvious commonality: Automation.

AWS leveraged automation internally to reduce the time to launch a new EKS version 243 days to 42, allowing them to launch five versions this year and catch up to upstream Kubernetes. Infrastructure Automation is also solving a diverse set of problems among AWS customers. For example,

• A large banking institution built common IaC modules with defined policies that could be deployed across multiple regions, saving their Saturday Dev teams 40 hours per month.
• HashiCorp Terraform was essential to migrating a 20TB file system to AWS with only two hours of downtime and made building a failback plan very easy.
• Platform teams remain in high demand and Terraform Cloud is enabling them to create self-service infrastructure in conjunction with the AWS Service Catalog, allowing their customers to efficiently get the resources they need.

Automation also applies in the security space. PKI remains a bottleneck, with some companies spending two months per year just managing certificates. Automating this process with HashiCorp Vault and AWS CertificateManager can eliminate this toil while increasing an environment's security posture.

Every success story I heard at AWS re:Invent featured a cloud environment that was designed around and built by automation, and our team is excited to assist our customers on this journey.

The role of partners in enabling GenAI

from Leo Hart

My key takeaway from the AI-themed re:Invent is relatively straightforward: Partners like WWT are poised to play a key role in helping customers keep security and privacy at the forefront of their GenAI journeys. Early missteps in addressing AI security and ethics can be detrimental to the successful long-term adoption of AI.

Partners that are successful in helping their customers mature their data and AI strategies will be able to adopt GenAI faster because, although GenAI is often considered separate, it's really part of an organization's data strategy. AI models don't differentiate customers. It's the customer's actual data and data delivery capabilities, architected correctly, that will enable AI-powered solutions and set them apart once an intelligent model is deployed.   

I learned that the best way to partner with AWS in this arena includes:

• Focusing on joint-end customers and accelerating the business outcomes.
• Bring highly skilled, super proficient resources to the table to address customer challenges.
• AWS and partner investments to upskill, train and develop competency can expand the relationship immensely. And putting the necessary resources in place to go to market together makes all the difference in determining success.

Conclusion

We only touched on a sampling of everything announced at AWS re:Invent 2023. Thanks for reading. 2024 is shaping up to be an exciting year for WWT and AWS!