Amazon re:Invent 2022 CEO Keynote Summary
I had the pleasure to attend my first ever Re:Invent this year and in this blog I have summarized the keynote from Adam Selipsky, CEO of Amazon Web Services, who was excited to show innovation. Re:Invent 2022 took place in Las Vegas from 28th November to 2nd December and was attended in-person by over 50,000 people.
Adam mentioned that Amazon is customer obsessed and as always, most of the features announced during his keynote come from the feedback of their customers.
Amazon is taking a leading stance on sustainability. “Amazon is the worlds largest purchaser of renewable energy. We are committed to powering our operations by 100% renewable energy by 2025 and we are already on target with 85% on our way there” quotes Adam. Amazon also commits to become water positive by 2030.
During this tough time, organizations should lean in to cloud inorder to save costs. Existing cloud customers including many Fortune 500 companies are already saving upto 30% or more costs with cloud. Adam says that If you’re looking to tighten your belt, cloud is the place to do it.
Cloud provides the flexibility to have on-demand capacity and helps customer innovate faster reducing the time to market. Innovation should be the driver and it can keep organizations upto date and give edge over their competitors. Even in uncertain times innovation is critical.
During his keynote, Adam announced features which I have classified across four domains — Data, Security, High Performance Compute and Next generation platform.
Data:
Adam talks about vast realm of data, data explosion. Managing scale and growth of data is going to be a challenge for every organization.Need right tools, integration, governance to generate insights.
1. Tools — AWS has services across databases, analytics and machine learning to explore the vastness and complexity of your data across your whole organization.
Data — AWS emphasis on data tools and analytics services. It has eight purpose build non-relational databases & give relational data engines including Aurora which is the fastest growing service in the history of AWS.
Analytics — Amazon also has the broad and deep set of Analytics Services so that users don’t have to compromise on cost, performance or scale. For large scale distributed framework, you can use EMR. For real time processing, you can use Amazon managed service for Apache Kafka. For structured data where you need superfast querying results, you can use RedShift.
Last year Amazon announced Serverless RedShift leaving OpenSearch as the only Analytics service that did not have a serverless option. At Re:Invent 2022, Adam announced serverless option for OpenSearch, a service that Amazon OpenSearch Service securely unlocks real-time search, monitoring, and analysis of business and operational data for use cases like application monitoring, log analytics, observability, and website search.
AI/ML — Adam mentioned that Amazon offers most complete set of Machine Learning and AI Services His keynoted referenced capabilities of Amzon SageMaker, a service used by tens of thousands of customers across the industries, Train models with billions of parameters, Support for leading ML Framework tools including TensorFlow, PyTorch, MXNet
It also has a Sagemaker Studio which is the first IDE for machine learning.
2. Integration — Having a wide set of tools poses integration challenges. There are quite a few organizations that shared feedback about the manual effort, complexity and undifferentiated heavy lifting involved in building E.T.L pipelines is painful. Amazon has a vision of having a Zero ETL future & in that future data integration is no longer a manual
With this being said Adam announced preview of Amazon Aurora zero-ETL integration with Amazon RedShift. It is a rear-real time analytics and ML on transactional data that consolidates data from multiple Aurora databases. Updates are automatically propagated and the best part is — it’s serverless!
Amazon also wanted to help customers doing analysis on Apache Spark which is supported in EMR, Glue, Sagemaker. Currently if you want to move spark query on RedShift, you have to move data to S3 or find, download & configure open source connector to RedShift. To solve this problem, Adam announced Amazon Redshift Integration for Apache Spark which allows user to run spark queries on RedShift Data from EMR, Glue, SageMaker within seconds
Adam commits to continue innovation in this space as a part of Amazon’s Zero-ETL vision
3. Governance — “To define full power of data, we need to make it easy for the right people and application to find access and share the right data when they need it” says Adam as he moves towards the data governance section.
Implementing data governance across organization is complicated & can depend from one organization to another. To ease out the data management for organizations, Amazon launched the preview of Amazon DataZone — Data management service to catalog, discover & govern data across the organization. It offers fine-grained controls to manage and govern access to data. Data catalog populated by ML is easy to search using business terms. Data Zone can be integrated with RedShift, Athena and QuickSight and provides APIs to third-party sources.
It offers fine-grained controls to manage and govern access to data. Data catalog populated by ML is easy to search using business terms. Data Zone can be integrated with RedShift, Athena and QuickSight and provides APIs to third-party sources
4. Insights — Once there is governance in place we want it easier for people to manage and explore data and we want to make discovering and sharing data insights lightning fast.
Amazon offers a service called Amazon QuickSight, a scalable, embeddable, ML-powered BI service build for the cloud. It’s most popular capability is Q which is a way of asking questions to data.
A night before Re:Invent 2022, Amazon announced Operational paginated reports for QuickSight and during Adam’s keynote he announced ML Powered forecasting with Q. This ML Powered forecasting with Q.
Adam says “We [Amazon] are investing across entire data journey from ingesting, storing, querying data and analyzing, visualizing and running machine learning and producing end to end governance to make it easier for customers to unlock the value of their data”
Security:
On AWS, building security is the path of least resistance and the reason why organizations rely on AWS because it provides the confidence to explore with security.
During this keynote Adam highlighted four reasons why customers choose AWS for security -
- Secure Global Infrastructure — Security is the central pillar around which services are designed and managed. AWS takes global infrastructure security seriously & today AWS is recognized as a highly security environment, passing the highest scrutiny
- Secure and Agile — Builders no longer have to trade-off security with speed. On AWS building securely is the path of lease resistance.
- Tools — Amazon offers a wide range of security focused services which have been built to help customers to identify and remediate vulnerabilities. Adam highlighted the capabilities of Amazon Guard duty, a service used to detect and respond to security threats. It continuously monitors your AWS accounts for malicious activity and unauthorized behavior. The advanced detection of guard duty uses ML, anomaly detection and threat intelligence feeds, detailed security findings for remediation
Earlier this year, Amazon released guard duty released guard duty for EKS protection that can help you see what’s happening outside the containers. It’s also important to see what’s happening inside the containers for which During this keynote Adam announced new Amazon GuardDuty capability of Container runtime threat detection. It will now detect threats running inside your containers - Provide access to broad marketspace tools that can be integrated — This covers use cases like application security, data security, data protection, perimeter protection, compliance
All log sources produce logs in their own unique frequently incompatible format, so to help define a standard data format for all of the security data. Security data is also usually scattered across your environment from application, firewalls and identity providers. To solve this challenge, Amazon announced Amazon Security Data Lake that automatically collects, combines and analyzes security data at petabyte scale
Computing:
Amazon also plans to emphasize on designing of high performance compute to help it’s customers reduce the time take to process and compute workloads. It offers broadest and deepest compute choice with over 600+ instance types for virtually every workload.
Graviton3. Amazon is on it’s 3rd generation of Graviton which is 25% faster than it’s second version. Amazon has over 100+ Gratiton instances but as per Adam, they are far from done. He went ahead and introduced the preview for C7gn instances for EC2 powered by Graviton.
Adam mentioned that there are also workloads that require extreme performance, one of them is Machine Learning. There are two types of ML workloads viz. training and inference. With rapid growth of data, the growth and model size is expanding the time required to train a model which drives exponential growth in training costs. Using Amazons TRN1 you can achieve 50% lower training costs.
Once you have trained your model to generate predictions of real time, it can require extreme performance for which we have build Inf1 instances powered by Amazon’s Inferentia chip to deliver low latency and low cost inference. To achieve higher throughput and latency, Adam announced Inf2 Instances for EC2 which would be powered by new Inferentia 2 chip. This enables the customer to get 4x the throughput and 1/10th the latency of in comparison with Inferentia 1 chips
Amazon also plans to emphasize on designing of high performance compute to help it’s customers reduce the time take to process and compute workloads. Currently Amazon customers have Hpc6a which offers best price for compute-intensive HPC. Depending on the size and complexity, many compute intensive workloads also require more network performance and for such workloads, a night before Adam’s keynote, Peter had announced HPC7g. HPC7g would be powered by elastic fabric adapter and Graviton 3E processors with upto with up to 35% higher vector instruction processing performance than the Graviton3
Adam did not stop here, for data and memory intensive workloads, he announced the Hpc6id for Amazon EC2.
Next Generation Platform:
Spatial simulation — Spatial simulation allows you to model high level entities like people, objects and how they act within a multidimensional environment. It requires millions of real time interactions. You can model urban cities like London or LA. Such simulations can be used to check responsiveness, disaster recovery and lot more. It can be an expensive thing and today if we are doing any such simulation you would want to use a 3D engine which are well-suited for virtual environment.
Adam announced AWS SimSpace Weaver to run massive spatial simulations without managing infrastructure. AWS SimSpace Weaver is a managed service that lets you create expansive simulation worlds at increased levels of complexity and scale. Multiple Amazon EC2 instances manage the underlying compute, memory, or networking so that you can devote time to building simulations — all while paying only a fraction of the cost of on-premises solutions.
Contact Center — Amazon aims to remove constrains and reimagine contact centers to make it more inexpensive and flexible. It offers Amazon Connect which is a contact center in the cloud that helps customer setup contact center in minutes and easily scale to support millions of customers. With an overview of the service, Adam went ahead and introduced 3 new capacities of contact center service -
- ML driven forecasting, capacity planning and scheduling
- Contact Lens with agent performance management
- Agent workspace with guided step-by-step actions. A new user interface that guides agents through customer interactions so that they can resolve issues faster
Supply Chain — As per Adam, Supply Chain in one of the other areas in which customers have been experiencing problems. Supply chain involve coordinating complex global network of suppliers, parts of manufacturing sites, distribution facilities & transportation providers — all while trying to deliver the right goods, at the right time, at a lower cost. The last two years have highlighted the importance of supply chain resiliency. Addressing SC issues around inventory is critical. Over the last 25 years, Amazon has solved a lot of supply chain complexity issues for themselves. Adam was thrilled to announce AWS Supply chain (preview) — cloud-based application that unifies data, provides ML-powered actionable insights, and offers built-in contextual collaboration. He gave a preview of dashboard and capabilities of the service.
Protected Environments for Analyzing data — One common problem across multiple industries is collaborating on shared datasets while still protecting the underlying raw data. Data clean room, protected environments where multiple parties can analyze the combined data without ever exposing the raw data that emerged as a solution. But such clean rooms are hard to build. The complex development can take months & once you build a clean room you have to main it throughout. To solve this problem, Adam announced a new service, AWS Clean rooms is a new service that makes it easier for customers and their partners to analyze and collaborate on their collective datasets to gain insights without revealing underlying data to one another. Adam emphasized AWS’ vision to make it more easier to incorporate more insights about customers and take relevant, targeted actions while protecting customer privacy.
AWS for Health — AWS is continuing to find ways where cloud can advance areas of healthcare & life sciences. Customer are examining all kinds of biological data. They are studying DNS or biological source code. They are also studying RNA, genetic molecules that help cells make proteins. They are also studying how these proteins interact together to improve drug effectiveness. The study of this molecular data is called omics. This includes familiar areas like genomics as well as lesser known ones like transciptomics and proteomics. Building systems to process omics data is challenging and often slows scientific discovery. It needs to be scalable, cost effective and capable of performing compute for millions of biological samples. This data should also be safe and with all these requirements in mind, Adam excitingly announced general availability of Amazon Omics. It is a purpose build service to store query and analyze, generate insights from genomic and other omics data. This is a service that can change the healthcare industry. Once your data is in Omics, it can be analyzed with Athena and EMR. It can also be integrated with Amazon HealthLake or Amazon SageMaker & comes with built-in access control and logging.
Reimagining retail industry — Adam highlighted the problems customer face in a store where the line is so long that customers want to bail. This was a very simple problem for everyone to understand and Adam immediately announced Just Walk Out, a technology by Amazon. It uses a combination of computer vision, sensor data & deep learning. A customer just scans the phone when they run into a store, take what they want and leave. No standing in lines to checkout anymore. Adam mentions that the company didn't stop there, they explored other places where people would stand in line like getting into stadium or badging into work. This led to Amazon One — a new simple way to identify, enter and pay with your palm!!! Cool isn’t it!
Adam’s last few minutes of the keynote on innovating in uncharted territory, to see and seize the opportunities. It all comes from constant feedback of the customers.
Personally for me it was a session with a lot of information to digest. Amazon continues to come up with features that can transform organization using the power of cloud. What’s your favorite feature announced during this Keynote and at Re:Invent 2022? Feel free to drop it in comments.