New Instances - Amazon EC2 Mac
Amazon EC2 Mac lets you run your apps on-demand on macOS in the cloud. Using AWS Nitro System, EC2 Macs are based on Apple Mac mini with Intel Core i7 processor (3.2GHz - 4.6GHz turbo, 6 physical / 12 logical cores and 32GiB of RAM) and operating systems: macOS Mojave (10.14) or macOS Catalina ( 10.15). macOS Big Sur (11.0) is also coming soon.
Developers can now build, compile, sign, test, and develop apps for iPhone, iPad, Mac, Apple Watch, Apple TV, and Safari for the first time in the AWS cloud, customizing and automating their development processes for Apple platforms.
More details here
New instances: R5b, D3 / D3en, M5zn, C6gn
The following new instance types are available today:
- R5b: 60 Gbps 260 . IOPS, 3 , R5. , . R5b Amazon Relational Database Service (RDS) , , Oracle Database SQL Server.
- D3 / D3en: HDD- 336 TB D2. , I/O . , D3 , HDFS MapR FS, D3en — Lustre, BeeGFS, GPFS, , Amazon EMR, Spark Hadoop.
- M5zn : Instances with the fastest Intel Xeon Scalable processors in the cloud, all-core turbo up to 4.5 GHz. M5zn instances are well suited for gaming, high performance computing (HPC), and simulation workloads such as finance, automotive, energy, and more.
The following instance type will be available later in December 2020:
- C6gn: Graviton2 ARM, 40% / x86. C6gn 4 , 4 2 EBS C6g. , , , HPC
AI/ML: AWS Trainium Habana Gaudi
AWS offers a wide variety of machine learning hardware, from NVIDIA and AMD graphics cards (with the new G4ad instance type) to their own AWS Inferentia chips that accelerate neural networks in production. AWS is introducing new AWS Trainium chips this year. AWS Trainium Chips accelerate neural network training and, together with AWS Inferentia Chips, enable faster and cheaper machine learning development, gaining benefits throughout their lifecycle.
Instances with Intel's Habana Gaudi accelerators will also be available next year. Such accelerators are effective in training neural networks, especially in the tasks of natural language processing, building recommender systems, and image / video processing.
More details here .
EKS Distro
Amazon EKS Distro is a Kubernetes distribution used to create clusters on Amazon EKS. It includes binaries and source code for components such as Kubernetes, etcd (a database that stores cluster configuration), as well as network and storage plugins. Customers can now install EKS Distro not only on AWS infrastructure, but wherever business needs require it, such as directly to physical servers in a data center “on the ground”, to VMware vSphere virtual machines or Amazon EC2 instances.
Clusters built on top of EKS Distro will support the same Kubernetes versions and required dependencies and patches as Amazon EKS so that customers can build reliable and secure Kubernetes clusters on their own without having to independently monitor new versions of various components and their compatibility. For notifications of new versions, just subscribe to the Amazon Simple Notification Service (Amazon SNS) topic. In addition, EKS Distro provides support for the Kubernetes versions supported by Amazon EKS, even if community support has ended for them.
More details here and here .
ECS Anywhere, EKS Anywhere (Announced: Planned in 2021)
Amazon ECS Anywhere and Amazon EKS Anywhere will be released in 2021, which will allow customers to run the ECS and EKS data plane, respectively, on their infrastructure, including their data center. At the same time, customers will be able to continue to use the same APIs for cluster management as for clusters fully deployed in the cloud.
Customers will be able to use ECS Anywhere and EKS Anywhere for workloads that cannot currently be migrated to the cloud. Only the data required to manage the cluster will be transferred to AWS, and if not connected to the cloud, workloads will continue to run.
Read more here (about ECS Anywhere).
Addons in Amazon EKS
Amazon EKS users can now install and manage add-ons on Amazon EKS clusters using the AWS Console, Command Line Interface (AWS CLI), and API. Amazon VPC CNI add-on management is currently supported . In the future, support for other addons will be released.
Kubernetes users often use add-ons for scaling, networking and security management, and monitoring. Previously, addons were managed manually directly in the Kubernetes cluster, now you can create a cluster with the necessary addons, add addons to an existing cluster or update installed addons directly through the EKS console without the need for additional actions.
More details here .
Amazon ECR Public
Amazon Elastic Container Registry (Amazon ECR) allows you to create fully managed private container image repositories. Customers can now also publish container images to a public directory using Amazon ECR Public, where they can be downloaded and used even without an AWS account.
Previously, developers needed to use multiple services to create private and public repositories. Both options are now supported in Amazon ECR, which provides highly available and scalable storage for container images, helm charts, and other OCI artifacts that are reliably replicated across two AWS Regions for faster load times and increased reliability.
More details here .
Lambda on containers + lambda 10Gb memory, 6vCPUs + 1 ms billing
Lambda on containers . Lambda applications can now be built as Docker containers up to 10 GB in size and published to AWS Lambda in the cloud. The base images for building their applications have been published, and the Lambda Runtime (Runtime Interface Clients (RIC)) will be published as an open-source project. This will make it easier to build applications using familiar tools and Docker cli, package all dependencies, and also make it much easier to test such lambda applications locally.
More details here
Lambda billing granularity reduced from 100ms to 1msThis will help reduce costs for scenarios where Lambda runtime can be very short and less than 100ms. For example, for dynamic web pages, the processing of which sometimes takes a couple of tens of milliseconds, the costs will drop by tens of percent.
More details here
Lambda functions can now use up to 10GB of memory instead of 3GB previously and up to 6vCPUs . This will make it easier to use lambda for tasks like video processing, ETL, Batch tasks, etc.
More details here
Aurora Serverless v2
Amazon Aurora is a MySQL and PostgreSQL compatible relational database built for the cloud. It runs 5 times faster than standard MySQL databases and 3 times faster than typical PostgreSQL databases. Amazon Aurora Serverless lets you run Aurora without the need for instance management and auto-scaling. Amazon Aurora Serverless v2, which is already available in pre-access mode for MySQL, improves auto-scaling and can change capacity in a fraction of a second to handle hundreds of thousands of transactions.
Amazon Aurora Serverless v2 provides customers with the full range of Amazon Aurora capabilities, including Multi-AZ, Global Databases, and Read Replicas. Thanks to improved automatic scaling, it is also suitable for large companies with thousands of applications with variable database load, as well as SaaS providers who do not want to run databases in multi-tenant mode, but want to have a separate cluster for each client without high costs and the need manual scaling.
More details here .
Babelfish MSSQL on Aurora (Preview) + Open-source project
Babelfish for Amazon Aurora is a new proxy layer for Amazon Aurora that allows you to make queries written for Microsoft SQL server. Babelfish understands wire-protocol and T-SQL, so you don't need to change SDKs or drivers to work with the database and rewrite applications.
This is a great opportunity to save on MS SQL license costs, and with AWS Database Migration Service (DMS) you can also migrate your data schema and data to Amazon Aurora as if it were an MS SQL server.
The Babelfish project will be available as an open source project later under the Apache 2.0 license.
More details here
Proton
AWS Proton is the first fully managed application deployment service for containerized and serverless applications. Platform teams can use Proton to connect and coordinate the various tools needed for infrastructure provisioning, code deployment, monitoring, and updates:
- AWS Proton integrates with widely used CI / CD systems and surveillance tools such as CodePipeline and CloudWatch.
- Provides curated templates that follow AWS best practices for general use cases such as web services running on AWS Fargate or streaming applications built on AWS Lambda .
- Gathers information about the deployment status of an application, such as the last date it was successfully deployed.
More details here
DevOps Guru
A fully managed operations service that enables developers and operators to easily improve application availability by automatically detecting operational problems and advising on how to fix them. DevOps Guru uses ML models on data from Amazon CloudWatch, AWS Config, AWS CloudTrail, AWS CloudFormation, and AWS X-Ray to identify anomalies:
- DevOps Guru automatically identifies operational problems, details possible causes, and recommends corrective actions.
- Visualize live data by integrating data from multiple sources that support Amazon CloudWatch, AWS Config, AWS CloudTrail, AWS CloudFormation, and AWS X-Ray.
- Collects logs and analyzes your environment. This can take up to several hours.
- A list of insights, which is a collection of anomalies that are generated when analyzing the AWS resources configured in your application.
- Aggregated metrics related to insight.
- Detailed graphs for each of the anomalies.
More details here
Amazon Lookout for Vision
New service in the category of computer vision for identifying visual defects in industrial products. Amazon Lookout for Vision helps you automate real-time visual inspection using computer vision and a trained model for your images with and without defects. About 30 images are enough. The best example of such work will be shown in the images below:
More details here
Amazon Panorama appliance
Now you can develop a computer vision model using Amazon SageMaker and then deploy it to a Panorama device, which can then run the ML model on video streams from multiple network and IP cameras. The Panorama device and associated console are currently in preview access.
- Add ML to existing cameras. Plug in your AWS Panorama device, connect it to your network, and the device will automatically discover your existing IP camera fleet.
- You can analyze video streams within milliseconds from a single control interface in multiple locations.
- Enable CV in environments with limited connections.
- Compliance with data privacy and governance requirements.
- Rapid development in a familiar programming environment.
More details here
Amazon Monitron
A comprehensive system that uses machine learning to detect abnormal behavior in industrial equipment, allows for predictable maintenance and reduces unplanned downtime. Consists of 4 key components:
- Amazon Monitron sensors. Easily install Amazon Monitron wireless sensors on rotating equipment such as motors, gearboxes, fans and pumps to measure vibration and temperature.
- Amazon Monitron Gateway. Data from Amazon Monitron sensors is automatically and securely transferred to AWS using Amazon Monitron gateways. Amazon Monitron gateways connect to sensors over Bluetooth Low Energy (BLE) and AWS over Wi-Fi.
- Amazon Monitron ML service. Sensor data is automatically analyzed using machine learning to identify abnormal equipment conditions that may require service.
- Monitron mobile application. Easily view sensor readings on the Monitron mobile app. The app sends push notifications when it detects abnormal behavior.
More details here
AWS Glue Elastic Views Preview
AWS Glue Elastic Views is a new AWS Glue feature that allows you to build materialized views from multiple sources using familiar SQL syntax, and then replicate the data to various services. It is a fully managed serverless service. Data is updated with minimal latency (near-realtime).
More details here
Amazon SageMaker Data Wrangler & Amazon SageMaker Feature Store
Data preparation and feature engineering take up 80% of the data scientist's work time. This work not only requires a lot of time and the use of many tools, but also in most cases of writing code.
Amazon SageMaker Data Wrangler can reduce your data preparation time from weeks to minutes. It automates data sampling and cleaning, feature engineering, and transformations. With Amazon SageMaker Data Wrangler, you can analyze and visualize data and evaluate the contribution of specific features to the accuracy of machine learning models.
Amazon SageMaker Feature Store helps you work consistently with features at all stages of the lifecycle - from data processing to training and deploying machine learning models to production. The Amazon SageMaker Feature Store provides a repository of features that can be used by a variety of applications and data scientists.
More details here and here
Amazon SageMaker Pipelines
Amazon SageMaker Pipelines is the first continuous integration and continuous delivery (CI / CD) system built specifically for machine learning. Amazon SageMaker Pipelines allows you to build pipelines that include data processing, model training and optimization, model deployment to production, model testing, and version control.
SageMaker Pipelines scale to fit the needs of the organization. Simultaneous work with thousands of experiments and hundreds of model versions is supported. Pipelines are built using the Python SDK and can be rendered in SageMaker Studio.
More details here
CodeGuru Reviewer Security + CodeGure Python support
Amazon CodeGuru Reviewer helps you find security bottlenecks and bugs in your code even before you deploy it. CodeGuru Reviewer Security Detectors will help you identify the main risk categories from the Open Web Application Security Project (OWASP), find places that do not reflect the best practices for building AWS APIs, and anything related to standard Java cryptography libraries.
Amazon CodeGuru now supports parsing code written in Python in addition to Java.
More details here
Amazon Connect Wisdom + Customer Profiles + Real-time Contact lens + Connect Tasks + Connect VoiceID
A whole series of updates and new functionality has been released for Amazon Connect. Amazon Connect is a virtual call center in the cloud that you can use to deploy your helpdesk or other services.
Amazon Connect Wisdom uses machine learning to help you find relevant and relevant information for an agent based on various data sources such as back-end systems, SalesForce, ServiceNow, various back-end wikis, knowledge bases, and more. The service tries to find answers to questions exactly in the form in which a person asks them.
Connect VoiceIDidentifies the caller by voice. At the first call, the client's speech, intonation, rhythm and other characteristics are analyzed. Next, a digital voice cast is created, on the basis of which, on subsequent calls, the conversation is verified in real time.
Real-time Contact lens - this extension allows you to analyze calls in real time and identify problematic telephone sessions when the voice rises or the sentimentality of the conversation and other indicators change greatly, which may require the intervention of a supervisor in the dialogue.
Amazon Connect Customer Profiles- this service provides a single customer card based on data from different systems, such as: Salesforce, ServiceNow, Zendesk, Marketo, etc. - which significantly reduces the time an agent searches for this information in disparate systems.
Amazon Connect Tasks can help you automate both routine tasks for agents and their actions in other systems. With Connect Tasks, you can assign tasks to agents based on their specialization or occupation.
AWS Small Outposts
New small AWS Outposts in sizes of 1U and 2U ( rack unit ) can now be easily placed in a store or office without the need to allocate space for an entire rack. Outposts 1U will be equipped with AWS Graviton 2 processors, and 2U Outposts server will be equipped with Intel processors. Both new form factors will support running EC2, ECS and EKS services locally.
AWS Russian Twitch Session
And now the promised announcement of the Russian-language twitch stream, which will take place on the key days of AWS re: Invent. Streams are prepared and conducted by leading AWS solution architects who have chosen all the most interesting and useful from the news and announcements of the multi-hour conference.
Register, connect to streams and discuss live .
AWS re: Invent. Main announcements of the first day (Part 2. Storage)