Amazon Re :Invent 2021 has just come to an end, and many developers are talking about this year’s blockbuster launch of Amazon Re :Invent.

This is not surprising. Every year, dozens of new products and features are released at Amazon Re :Invent, some of which may represent the future development direction of cloud computing and the entire IT infrastructure, such as Amazon RedShift and Amazon Lambda. The former has led to the development of cloud native data warehouse in the industry, while the latter has brought serverless to the attention of developers in the industry.

However, This year’s Amazon Cloud Technology Re :Invent is a bit different. In addition to regular iterations in product performance, it is more important to reflect the conceptual extension of cloud computing services. The most typical examples are Amazon IoT TwinMaker and Amazon IoT FleetWise.

The meta-universe and the Internet of Things: The world really is going cloud

According to Amazon IoT TwinMaker, developers can easily and quickly create real-world digital twins such as buildings, factories, industrial equipment and production lines. With Amazon IoT TwinMaker, users can use digital twinnings to build real-world applications that improve operational efficiency and reduce downtime.

A digital twin is a virtual mapping of a physical system that can be updated periodically based on the structure, state, and behavior of the real-world objects it represents. Amazon IoT TwinMaker makes it easy for developers to assemble data from multiple sources, such as device sensors, cameras, and business applications, and combine that data to create a knowledge graph that models real-world environments.

Digital twinning technology was first used in the health maintenance and security of aerospace vehicles, which is a cold technology. However, with the rise of the concept of “meta-universe”, digital twin technology is more and more well-known, because the essential feature of digital twin is the equivalent mapping of the physical world in the information world, so it has become one of the important supporting technologies of the concept of the meta-universe, especially the industrial meta-universe, is widely used.

Amazon IoT TwinMaker is as much about the industrial Internet as it is about the metasexes. Previously, the scope of cloud computing services, focusing on the Internet industry, for the so-called traditional industry, mainly to provide transformation services. But now, cloud computing is rapidly expanding its service outreach to include virtual mapping of the entire physical world through the bridge of the metasverse.

Closely related to this is the Amazon IoT FleetWise. Amazon IoT FleetWise enables automakers to easily collect and manage data in any format in their vehicles — regardless of brand, model, or configuration — and standardize data formats for easy data analysis in the cloud. When the data is in the cloud, automakers can apply it to a vehicle’s remote diagnostic programs to analyze the health of their fleet, help automakers prevent potential recalls or safety issues, or improve technologies like autonomous driving and advanced assisted driving through data analytics and machine learning.

If Amazon IoT TwinMaker provides mapping services from the real world to the virtual world, then Amazon IoT FleetWise focuses on the Internet of vehicles and solves the long-term development problems of the Internet of vehicles. The concept of the Internet of cars first emerged in the 1960s, but for 60 years it has been something of a limp development — most people think of it as providing Internet services in the car, rather than uploading vehicle data for analysis.

It was not until the launch of Tesla Model S in 2012 that the Internet of vehicles was incorporated into the assembly line of automobile production as a mandatory function. Now Amazon IoT FleetWise has been released, bringing the Internet of vehicles related services to the cloud.

A key feature of Amazon IoT FleetWise is the ability to build virtual representations of vehicles in the cloud and apply common data formats to build and tag vehicle attributes, sensors, and signals. Amazon IoT FleetWise standardizes vehicle modeling using the Vehicle Signal Specification (VSS) so that signals such as “fuel pressure” are always expressed as fuel pressure and measured in pounds per square inch (PSI) and kilopascals (kPa). After the vehicle is modeled, upload a standard CAN database (DBC) or AUTOSAR XML (ARXML) file so that Amazon IoT FleetWise CAN read unique proprietary data signals sent over the vehicle controller LAN Bus (CAN bus).

Amazon IoT TwinMaker and Amazon IoT FleetWise have the same underlying concept of building virtual mappings in the cloud, but one is for the industrial sector and the other is for the automotive industry. The world is virtualizing, as it were, and going cloud at the same time.

Amazon SageMaker Canvas: Create ML models without code

If Amazon IoT TwinMaker and Amazon IoT FleetWise are conceptual extensions of cloud services horizontally, Amazon SageMaker Canvas is conceptual extensions of cloud services vertically.

Everyone knows Amazon SageMaker as a four-year-old fully hosted machine learning service. Amazon SageMaker provides a complete “central kitchen” for developers, using Amazon SageMaker developers can just prepare the “ingredients” (data) and start cooking directly (training model), It dramatically increases the efficiency with which developers and data scientists can build, train, and deploy machine learning models, ushering in a whole new era of intelligence.

But the AI field has long been limited by talent shortages, and the number of applications for AI is increasing, so the threshold for machine learning services needs to be lowered further. This is exactly what Amazon Cloud technology is aiming to do with the release of Amazon SageMaker Canvas — to build machine learning models with no code, to make model predictions, and to ensure that services can be delivered without the help of data engineering teams. It uses the same technology as Amazon SageMaker to automatically clean and combine your data, creating hundreds of models behind the scenes, selecting the ones with the best performance, and generating new individual or batch predictions. Support binary classification, multi-class classification, numerical regression, time series prediction and other problem types.

There was a lot of debate about low code, no code, but now it seems that this is not a conceptual battle, but a real need in the industry. The launch of Amazon SageMaker Canvas demonstrates this.

At the macro level of the whole AI, no matter the prediction service or analysis service based on AI, it also breaks away from the pursuit of higher level ARTIFICIAL intelligence, and takes into account the ability of AI to empower the industry. This is the further expansion and implementation of the cloud computing concept of service.

Amazon Private 5G: Connect to IoT devices with proprietary 5G

In this context, the launch of Amazon Private 5G has attracted a lot of attention, as it is an important and necessary attempt to support the expansion of services. Amazon Private 5G is arguably one of the most important releases of Re :Invent.

On the mobile end, we already have 5G communications services, but what companies need is proprietary 5G service networks. Amazon Private 5G automatically sets up and deploys networks and expands capacity on demand to support more devices and network traffic. “With Amazon Cloud Technologies Private 5G, we extend hybrid infrastructure to customers’ 5G networks to simplify, quickly and cheaply build private 5G networks,” said David Brown, VICE president of Amazon Cloud Technologies EC2. Customers can start small, scale on demand, pay on demand, and monitor and manage their network from the Amazon Cloud Tech console.”

Amazon Private 5G also focuses on serving the huge sensor and end-to-end device cluster dominated by industry 4.0. The industrial meta-universe and Internet of vehicles mentioned above are also in the same sequence.

Koch, the largest unlisted company in the world, has reached a cooperation with Amazon Cloud Technology on Amazon Private 5G. Koch enterprise Group, whose core is petroleum and chemical industry, is also a representative service case of Amazon Cloud Technology.

Amazon Graviton3: The underlying computing power has been upgraded

Of course, both IoT services such as Amazon IoT TwinMaker and Amazon Private 5G rely on the performance of chips in the underlying instances. At this year’s computing conference, pingtou, a semiconductor company owned by alibaba, unveiled the yitian 710, its own cloud chip, and announced that its performance beats the Amazon Graviton2 by 20 percent.

From Amazon Graviton 2019 to Amazon Graviton2 2020 to Amazon Graviton3 today, Amazon GravitonTechnology has improved computing services from the chip. Compared to Amazon Graviton2, Amazon Graviton3 integrates 55 billion transistors for more than a 25% improvement in single-core performance and twice as much floating-point and encryption performance. On the machine learning side, Amazon Graviton3 includes support for Bfloat 16 data and will be able to deliver up to 3x performance. Along with the performance leap, the Amazon Graviton3 uses 60 percent less energy than its predecessor.

The new Amazon EC2 C7g instance is powered by the Amazon Graviton3 processor and is the world’s first on-cloud instance to support DDR5 content. 25% performance improvement over current generation Amazon C6g instances powered by the Amazon Graviton2 processor.

Of course, the Amazon Graviton3 is a general-purpose chip, and the dedicated chip has also been updated. Amazon Cloud Technology announced that the new Amazon Trn1 instance, powered by Amazon Trainium, Amazon’s second machine learning chip, will provide the best value for money for training deep learning models in the cloud for natural language processing (NLP), computer vision, search, recommendation, ranking and other use cases. Compared with the P4d instance, the cost of training the deep learning model through Amazon Trn1 instance is reduced by 40%.

Amazon Nitro System chips have also been released. Amazon Nitro is a super black tech. It’s an architecture that maximizes the amount of resources a server can provide to users and reduces the cost of virtualization. The so-called “virtualization loss” is the inevitable cost of network, storage, management and other system functions in order to maintain the normal operation of the service in the past, which accounts for 30% of the overall performance of the server. Nitro architecture focuses on these 30 percent of performance issues by customizing hardware.

The Amazon Nitro System chip, which supports the Amazon EC2 Instance underlying management platform, can share the workload for the CPU. General purpose chips, inference chips, Amazon EC2 support chips, this release is a clean SLATE. The Im4gn/Is4gen/ I4i instance of Amazon Nitro SSDs provides 30 TB of NVMe storage, with a 60% reduction in I/O latency and 75% reduction in latency variability compared to the previous generation I3 instance.

Data Serverless: Rapid promotion of Serverless applications

Of course, in addition to the expansion of service capabilities at the network and end levels, this update at the Serverless level is also noteworthy.

The industry knows that Amazon Lambda started the era of Serverless, but it was in 2019 that Amazon Lambda really gained widespread acceptance and followership in the industry. At Amazon Re :Invent, Amazon Cloud Technologies launched Serverless releases of four core products: Amazon Redshift Serverless, Amazon EMR Serverless, Amazon MSK Serverless, and Amazon Kinesis Data Streams On-demand.

Amazon Redshift, as we’ve mentioned, is one of the first cloud native repositories; Amazon EMR is a Hadoop hosting service provided by Amazon Cloud Technology. Amazon MSK is a Kafka hosting service; Amazon Kinesis Data Streams On-demand is a streaming data processing platform.

The Serverless versions of these services allow consumers to run applications built using these frameworks with a few clicks without the need to configure, optimize, or secure the cluster.

Cloud big data architecture, due to the Serverless version of Amazon Cloud Technology services update, its threshold is rapidly lowering. In the past, the construction of such architecture as intelligent lake warehouse made architects and engineers very headache, but now, the work of engineers is becoming simple adjustment — Serverless’s change to industrial ecology is almost permanent.

Write in the last

According to amazon Re :Invent 2021 and the release of cloud computing conferences this year, product changes in the cloud computing field focus on the upgrading of underlying basic computing power, which is a competition of core strength. Secondly, it pays attention to the extension of the outer edge of the service. How to understand the three roles of cloud, network and terminal, and provide universal public cloud services as far as possible, becomes the key. As a newly emerging concept, the meta-universe provides new ideas and directions for the overall development of technology, which is worth our special consideration.