Think in Context: AWS re:Invent 2020 Andy Jassy Keynote

(Translation in progress…)


Last year (finally) it was my first trip to participate in AWS re:Invent 2019. I arrived with couple tens of thousands of people (squeezed together) in Las Vegas to enjoy a full week of high-density information flow, which is very unforgettable.

With the experience of last year, although this year has been adjusted to be online, I can catch the pace and rhythm relatively, and can relax and enjoy the information bombing of the weeks in accordance with the Asian time zone (which year did not bomb?)

The number of Keynote sessions of AWS re:Invent this year (2020) has also increased and adjusted compared to last year. Last year’s show was the Monday Night Live, where everyone just started the carnival on Monday night. Peter DeSantis brought information about the infra. This year it became independent and directly renamed “Infrastructure Keynote”. This year also added a new “Machine Learning Keynote” brought by Swami Sivasubramanian.

The three remaining keynotes are “Andy Jassy Keynote”, “AWS Partner Keynote” and “Werner Vogels Keynote”. Each keynote has its own main axis. This article shares the Taiwan time zone. The rebroadcast of “Andy Jassy Keynote” starts at 8 am here. At this time, breakfast and a pot of hot tea are good fit. Don’t worry that it’s a few hours away from the first broadcast. AWS CEO Andy Jassy mentioned in this Keynote “Still early days for cloud”, and now the cloud only accounts for 4% of all IT expenditures, there is no problem with a few hours of shift.

For the full text, I tried to capture the structure of the speech, and then put some observations and inferences, and then put some running notes in each paragraph to facilitate future search and use. New services or new features are marked with [NEW 🚀] in this article, so that you can press Command/Ctrl+F to search inside the page. This article deliberately removes most of the product links first, so that everyone can focus on reading (we are less focused these years, right?). You are also welcome to give me some feedback or corrections.

Let’s get started!


這次 Andy Jassy Keynote 三個小時的演講架構,猛然一看會覺得超多東西很難吸收,但其實就是一個主題以及五個段落。整場演講前後加上開場、收尾,中間搭配客戶案例分享影片來做串場銜接。


  • Compute
  • Data Stores
  • AI
  • Challenges & Applications of Industries
  • Hybrid

五個段落首要兩個「Compute」和「Data Stores」是計算機的基礎設施,不論計算機位於地端或雲端、或特定場域、任何產業,都會遇到「Compute」和「Data Stores」。後面三個段落「AI」、「Industries」、「Hybrid」則是站在「Compute」和「Data Stores」的基礎上做延伸,做深,也做廣。符合一個主題「Reinvent」八個要點的第七點所提到的「the platform with the broadest and deepest set of tools」。順序大致是由硬而軟,然後再海陸雙拼

演講過程中,除了一個主題「Reinvent」作為核心之外,另外搭配第四個要點「Solving real customer problems」貫穿全局,也符合母公司 Amazon growth flywheel 的精神。


一些統計數字,搭配精簡的版型,幫大家暖暖身。(等一下就要拿很多很多很多 company logos 和架構圖轟炸大家囉 XDD)(客戶公司 logos 滿到一個長形螢幕擺不下,還加了個次螢幕作輔助,畫面效果超棒!)這個開場,個人覺得很重要的數字是在講「雲還處於初期 (Still early days for cloud)」這個概念,4% 這個數字代表著 AWS 不能放鬆警戒,後面的 96% 隨時可能會被翻盤。

對 AWS 客戶、學習及使用 AWS 的我們來說,如果 4% 這個數字所在的當下這個時期,我們就已經覺得 AWS 推出的 200+ 種產品有點多,學都學不完了。換另一個角度想,那 96% 在地端的 IT 支出往哪兒去了?平常有接觸到嗎?是被層層疊疊蓋掉了嗎?有文件嗎?有梳理過嗎?有 API 可以使用嗎?有效率嗎?還是不敢揭開的歷史遺機?

  • 3 weeks
  • 5 keynotes
  • 500,000+ people registered
  • $46B+ Revenue run rate (Annualized from Q3 2020)
  • 29% YoY growth, Q3 2020 vs Q3 2019
  • State of the cloud
    • 45%: AWS
    • 17.9%: Microsoft
    • 9.1%: Alibaba
    • 5.3%: Google
    • 2.0%: IBM
    • 20.7%: Other vendors
  • Total IT Spend (Still early days for cloud)
    • 4%: Cloud
    • 96%: On-premises
  • COVID-19 is pushing companies toward the cloud.

0. Reinvent

Andy directly disassembled “Reinvent”, which is the same name of this event re:Invent, into 8 points.

  1. The leadership will to invent and reinvent. (It’s hard to get the data.)
  2. Acknowledgement that you can’t fight gravity.
  3. Talent that’s hungry to invent. (When someone is retiring…)
  4. Solving real customer problems with builders.
  5. Speed.
  6. Don’t “complexify”.
  7. Use the platform with the broadest and deepest set of tools.
  8. Pull everything together with aggressive top-down goals.

Guest: Lori Beer, Global CIO, JPMorgan Chase

  • Funded Thomas Edison’s first lightbulb
  • Early enterprise ATM banking
  • broadest and deepest set of tools
  • Amazon SageMaker
  • Amazon Redshift

Guest: Jerry Hunter, Snap Inc. Senior Vice President, Engineering

Jerry used a Snapchat video to present it, which is fully in line with the concept of their own brand, which is awesome! XDD

  • Cloud native
  • Graviton2, Serverless
  • Lower cost, improved performance
  • "You might think that there isn't much left to reinvent when it comes to compute, but the innovations just keep coming."

1. Reinventing Compute


三個種類分別是「Intances」、「Containers」和「Serverless」,接著因為不同團隊、部門之間選擇了不同的種類來實作,造成了一些流程上的整合挑戰,因此提出了一個整合流程方案「AWS Proton」。

Compute Types

  • Intances
  • Containers
  • Serverless


  • Amazon EC2
    • [NEW 🚀] Largest local storage instances with D3en (336TB)
    • [NEW 🚀] Only cloud provider with on-demand macOS instances
    • Only cloud provider that supports Intel, AMD, and Arm processors.

「We’re reinventing virtualization with Nitro.」

接著聊使用 Nitro 來增進虛擬化技術,準備帶出第二個種類「Containers」。但在開始進到 Containers 之前,先介紹超棒性價比的 Graviton2 (ARM processer) 、以及關於 machine learning 的 AWS Inferentia。

AWS 推出的 Graviton2 (ARM processer) 是一種 EC2 instance type,讓我們可以選擇在 ARM 處理器架構上運行程式。這一年來越來越多 Docker images 都陸續支援多處理器架構,例如 Docker Hub 上的 php images 就支援了 linux/386, linux/amd64, linux/arm/v5, linux/arm/v7, linux/arm64/v8, linux/mips64le, linux/ppc64le, linux/s390x 八種處理器架構。我也正在進行這類處理器運行 PHP 的效能比較,有興趣的話歡迎追蹤 image:dwchiang/nginx-php-fpm 或我的 Twitter @dwchiang

Andy 也引用了 (Liz Fong Jones) 和 NEXTROLL (Valentino Volonghi) 的推文,來點出 Graviton2 (ARM processer) 所帶來的性價比。(接下來就來敲碗 Fargate 和 Lambda 也能支援 ARM 架構囉!會不會支援 M1 呀?:p 以成本來說應該是會主推自家開發的 Graviton 系列處理器,但也許客戶的聲音說需要 M1,M1 就會來了 XDD)

  • Graviton2 (ARM processer) - R6g, M6g, C6g, C6gn, T4g - 40% better price/performance for all workloads - (Liz Fong Jones): - If M6g were the only instance type in our fleet of Shepherds, we could run 30% fewer instances in total, and each instance would cost 10% less on-demand versus C5. - Saving 40% on the EC2 instances bill for this service once we’re able to fully convert our instances to Graviton2 is well worth the investment. - NEXTROLL (Valentino Volonghi): - Almost 50% cost saving with Graviton2.

接著針對 machine learning training,將於 2021 年上半年推出 Amazon EC2 instances powered by Habana Gaudi,一種新的 EC2 instance type,採用來自 Intel 旗下 Habana Labs 的 Gaudi 加速器,可讓目前 GPU-based EC2 ML training instances 增進 40% 性價比。

  • Lowering the cost of machine learning inference with AWS Inferentia. - Alexa has reduced their cost of inference by 30% and lowered their latency by 25% using our Inf1 instances… - [NEW 🚀] Habana Gaudi-based Amazon EC2 instances (Available first half of 2021) - Up to 40% better price/performance over current GPU-based EC2 ML training instances. - [NEW 🚀] AWS Trainium (Available in 2021) - Our ML training chip custom designed by AWS to deliver the most cost-effective training in the cloud. - Most teraflops of any ML instance in the cloud - Support for TensorFlow, PyTorch, and MXNet - Uses same Neuron SDK as Inferentia - Available as EC2 instances or in Amazon SageMaker


隨著容器技術在全世界各產業陸續導入,這個階段主要想要解決的問題是如何讓容器可以自在地在天地之間移動或部署,由此而生的 Amazon ECS Anywhere 與 Amazon EKS Anywhere,並且將 EKS Distro 給開源 open sourced 了。

在 Youtube 上 Massimo Re Ferre (Principal Technologist at AWS) 也馬上釋出了 30 分鐘的 Amazon ECS Anywhere Demo 影片,可以在兩台 Raspberry Pi 上安裝 ECS Agents 後成為 ECS container instances,然後建立 ECS Cluster、執行 ECS Tasks,超棒的!延伸閱讀請參考 Massimo Re Ferre 的「Introducing Amazon ECS Anywhere」一文。

Amazon ECS/EKS Anywhere 推出後,可以將地端開發環境與雲端量產環境的部署流程整合成一套,也可以創造更多接地氣的應用場景得以實現。Amazon ECS Anywhere 實作的方式是新增了第三類的「EXTERNAL Launch Type」,我於 2020/12/04 AWS DEV DAY TAIPEI 2020 會分享「技術選型:Amazon ECS Launch Types:Fargate vs. EC2」,待我實測「EXTERNAL Launch Type」後,也打算來列入比較與各位分享。

關於更多 Amazon ECS,歡迎延伸閱讀 Ernest 的 學習筆記:Amazon Elastic Container Service (Amazon ECS)

  • The units of compute keep getting smaller. - AWS is the best place to run and manage containers: EKS, ECS, Fargate. - 三個產品線都持續成長中。同一個客戶的不同團隊採用不同方案。 - The management challenge of moving containers to the cloud. - [NEW 🚀] Amazon ECS Anywhere (Run ECS in your own data center!) - [NEW 🚀] Amazon EKS Anywhere (Run EKS in your own data center!) - [NEW 🚀] Open sourced EKS Distro


AWS Lambda 身為 Serverless 的一環,這次帶出兩個亮點,一是收費方式改變,最多可節省 70%!另一是直接 GA 讓大家玩 Lambda Container Support!

  • AWS Lambda - [NEW 🚀] 1 millisecond billing lowers costs up to 70% (vs. 100 ms billing) - [NEW 🚀] Lambda Container Support (GA TODAY!) - Build Lambda-based applications using existing container development workflows. - Package code and dependencies as a Docker or OCI compatible container image. - Use a consistent set of tools for containers and Lambda-based applications. - Deploy Lambda functions built on top of third-party base container images.


  • The challenge of moving to smaller units of compute. (應該是地端、IoT 要放量了?!大家快上!XDD) - monolith vs. microservices - 跨團隊之間很難整合部署流程 - [NEW 🚀] AWS Proton
  • AWS is changing the game for what customers can build. - 10 年前我們大家都還沒開始談論 serverless, container.

邀請嘉賓 Don MacAskill (CEO and Chief Geek, Smugmug and Flickr) 開場後,主題即將進入資料存儲。

Guest: Don MacAskill, CEO and Chief Geek, Smugmug and Flickr

  • 在 AWS S3 剛發佈 2006 那年就將 6 TB 的資料搬進來了!(Andy 還強調不是 6 GB,是 6 TB!)
  • 當年的案例影片可以參考這裡,還可以看到當年的 AWS 網頁。
  • “The cloud has totally reinvented how we store, secure, analyze, and share data at a scale that we coun’t have even imagined 14 years ago.”

2. Reinventing Data Stores

還記得剛開場 Andy 提到的「Speed」嗎?速度這不就來了?:p

  • 超多資料被產生 der!!
    • More data is created every hour today than in an entire year 20 years ago.
    • More data will be created in the next 3 years than in the prior 30 years combined.
  • Block storage is fundational and one of the most pervasive forms of storage. - Amazon EBS, gp2 is the general purpose volume type - [NEW 🚀] gp3 volumes for EBS: 4x peak throughput (GA Today!) - io2, 4x more max IOPS than gp2 and gp3 - [NEW 🚀] io2 Block Express, first SAN for the cloud! More SAN features coming in 2021. (Available Preview Today!)

大家都想逃離腳邊或地端的資料庫…(歷史遺機再次出現… 喔不,這不是歷史,他們還在線上!)

  • On-premises databases require a lot of undifferentiated heavy lifting. - “Incumbent” relational databases - Amazon Aurora - Fastest-growing AWS service, ever - MySQL and PostgreSQL compatitable - Serveral times faster than standard MySQL and PostgreSQL - Highly available and durable(延伸閱讀 Ernest 筆記 AWS Product List: Availability/SLA) - 1/10th the cost of commercial-grade databases - Amazon Aurora Serverless - Automatically starts up, shuts down, and scales capacity up or down - No database instances to manage - Scales database capacity to meet demand within 5-50 seconds. - [NEW 🚀] Amazon Aurora Serverless v2 (and reinvented) (Available in Preview Today!) - Scale to hundreds of thousands of transactions in a fraction of a second. - Scale capacity up and down in fine-grained increments for just the right amount the application requires. - Save up to 90% vs. provisioning capacity for peak load. - Multi-AZ support, Global Database, Read Replicas, Backtrack, and Parallel Query.

打完祭司關卡後,接著打有點軟的關卡!有在使用 SQL Server 的應用程式有機會不需修改程式就直接與 Babelfish for Aurora PostgreSQL 合體營運!

  • More than 350,000 databases migrated to AWS. - AWS Schema Conversion Tool (SCT) to migrate database schema - AWS Database Migration Service (DMS) to migrate data with minimal downtime - [NEW 🚀] Babelfish for Aurora PostgreSQL (Available in Preview Today!) - Run SQL Server applications on Aurora PostgreSQL with little to no code changes - Stop paying for SQL Server licenses you don’t need - New translation capability to easily run SQL Server applications on Aurora PostgreSQL. - Understands SQL Server’s propreietary dialect (T-SQL) and communications protocol (TDS). - Migrate the data with DMS, then update your application configuration to point to Aurora instead of SQL Server. - [NEW 🚀] Babelfish for PostgreSQL, open source project! Available on GitHub in 2021.
  • "When all you have is a hammer, everything looks like a nail."
  • The right tool for the right job.

有了這麼多的資料,那… 應該可以拿來分析,對吧?

是的,接下來 Andy 哥將介紹一個關於資料分析的新工具來解決資料分析時的一個痛點:Data movement。

  • Data movement! - 資料圍繞著 Amazon S3 被搬來搬去、移來移去。 - [NEW 🚀] AWS Glue Elastic Views (Available in Preview Today) - Easily build materialized views that automatically combine and replicate data across multiple data stores.

到此 data stores 告一段落,截至目前為止已經用掉了將近 50% 的演講時間,說「Compute」和「Data stores」是整個計算機的兩個核心,這比例抓得頗恰當。

接下來看一段來賓所帶來超炫的造飛機影片,而且是超音速飛機,話題也將從 data stores 前進到 AI 等應用。

Guest: Black School, Founder and CEO, Boom

  • 用 AWS 設計飛機
  • 53 million core hours –> 100 million core hours
  • 有 JAL 就給讚!XDD

(讓我歪樓一下,多放些飛機的照片 :p)

Guest: Marianna Tessel, Intuit CTO

  • "Intuit is an early adopter of AI, starting the journey over a decade ago. We see tremendous opportunity in applying AI to revolutionize our business and benefit out customers. And... this is just the beginning."

3. AI

  • Framework used for new machine learning scientific publications
    • 9 out of 10 ML practitioners use more than one framework
    • 60% of ML practitioners use more than two frameworks

ASK #1: Need the right tools for expert ML practitioners

  • The broadest and most complete set of ML capabilities

ASK #2: Tools for everyday developers and data scientists

  • Data preparation for ML is hard!
    • Write queries and code to download raw data from data stores
    • Figure out and prototype the conversion, transformations, and feature combinations.
    • Transform the data, spin up infrastructure to run code, monitor, manage for each change, validate, and save the results.
    • Make sure it all works - data point by data point - and fix any errors, outliers, or missing data points - [NEW 🚀] SageMaker Data Wrangler (GA Today!) - The fastest way to prepare data for ML
  • Storing, saving, and reusing ML features is hard!
    • Need to be able to maintain variations and combinations of features for many different models
    • Need to build and share features across multiple teams
    • Need to be made available for inference in real time
    • Need to ensure consistency of features for both training and inference - [NEW 🚀] SageMaker Feature Store (GA Today!) - A new repository that makes it easy to store, update and share ML features. - Purpose-built and accessible from SageMaker Studio - Easily name, organize, find, and share features - Access features in batches or subsets - Low latency for inference
    • [NEW 🚀] SageMaker Pipelines (GA Today!)
      • The first purpose-built easy-to-use CI/CD service for ML
      • Define each step of your end-to-end ML workflow
      • Pre-configured, customizable workflow templates
      • Logs each step in SageMaker Experiments
      • Workflows can be shared and reused

Ask #3: ML for those who don’t want to build models

  • Amazon CodeGuru
    • CodeGuru Reviewer provides automated code reviews for Java - [NEW 🚀] Amazon CodeGuru for Python - CodeGuru Profiler identifies the most expensive lines of code - [NEW 🚀] CodeGuru Security Detector provides real-time alerts when developing code that may not be secure
  • [NEW 🚀] Amazon DevOps Guru (Available in Preview Today)
    • A new service that uses ML to identify operational issues long before they impact customers.
      • Missing or misconfigured alarms
      • Early warning of approaching resource limits
      • Code and config changes that may cause outages
      • Under-provisioned compute capacity
      • Database I/O overutilization
      • Memory leaks

ASK #4: Not having to figure out how to put pieces together

  • Hire you to get things done. (No metter using which technology.)
  • Reinventing business intelligence
  • Amazon QuickSight
  • [NEW 🚀] Amazon QuickSight Q (Available in Preview Today) - Ask Q any question in natural language and get answers in seconds

4. Challenges & Applications of Industries

Guest: Paul Cheesbrough, CTO and President of Digital, Fox Corporation

  • “Working with AWS, we have reinvented how we produce and distribute content to our customers across all platforms and devices using the cloud.”

Traditional Contact Centers

在疫情期間 Contact center 的需求上升,在這領域加強功能更能貼近與滿足客戶需求。從客戶需求的脈絡來理解 AWS 推出的對應產品會輕鬆一些 :)

The challenges of traditional contact centers.

  • Amazon Connect: Contact center in the cloud - The same customer service technology used by Amazon - Built from the ground up with the cloud and AI/ML in mind - Set up and configure a contract center in minutes - Easy to use and configure for non-technical users - Scale from tens to tens of thousands of agents - Save up to 80% over traditional contact center solutions - No infrastructure to deploy or manage
  • What we heard from customers - Agents need the right info at the right time. - How can we get the right data about products or service provided to agents when a customer is inquiring? - How can we get data on a customer’s related activity and experiences? - [NEW 🚀] Amazon Connect Wisdom (Available in Preview Today) - A new capability that uses ML to deliver agents the product and service information they need to solve issues in real time - [NEW 🚀] Amazon Connect Customer Profiles (GA Today!) - Gives agents a unified profile of each customer to provide more personalized service during a call - We want to identify and react to customer issues in real time - Can you make it easier to understand and react to customer contacts that are going sideways… and do so in real time? - Contact Lens for Amazon Connect: Contact center analytics for Amazon Connect - [NEW 🚀] Real-Time Contact Lens for Amazon Connect (GA Today) - Help us optimize customer service agents' time - How can we help agents manage their time outside of calls? - How can we optimize agents’ time on the phone? - [NEW 🚀] Amazon Connect Tasks (GA Today) - Automates, tracks, and manages tasks for contact center agents - Connect Tasks makes follow-up tasks easier for agents and enables managers to automate some tasks entirely - [NEW 🚀] Amazon Connect Voice ID (Available in Preview Today) - Real-time caller authentication using ML-powered voice analysis - Connect Voice ID provides real-time caller authentication without disrupting natual conversation

What industriies are being reinvented?

Reinventing automotive

  • Rivian: Cut the time of each crash simulation test by 55%, from 18 hours to 8 hours.
  • TOYOTA: Managing connected vehicles around the world.
  • BlackBerry: Intelligent Vehicle Data Platform, IVY
  • BWW GROUP: Develoment of safe and performant automated driving systems
  • kyft: Gather petabytes of data to improve self-driving system
  • VOLKSWAGEN GROUP: Connecting data from 120+ facilities, 1,500 suppliers, and 30,000 locations.

Reinventing healthcare

  • PHILIPS: Analyzing petabytes of data to diagnose disease faster
  • Cerner: Predicting patient risk with ML models
  • moderna: Reduced COVID-19 vaccine development time from 20 months to 42 days
  • Mount Sinai: Creating an integrated patient experience

Reinventing media and entertainment

  • FOX: Single platform to deliver broadcast and on-demand content
  • VIACOMCBS: Movin all of their media operations (including broadcast) to the cloud
  • Disnep: Sacled Disnep+ to over 70M subscribers in less than a year
  • COMCAST NBCUNIVERSAL: Produce content remotely - spun up hundreds of virtual media workflows
  • NETFLIX: Tools for other content producers to produce higher-quality content faster

Reinventing manufacturing

Guest: David Gitlin, President & CEO, Carrier

  • 100 year old start-up
  • Focus: World leader of healthy, safe and sustainable building and cold chain solutions.
  • 做了冷鏈保護食物,發現有一大堆食物被浪費了。(咦?

Data is the connective tissue for industrial processes

  • Improving industrial processes - 1. Machine Data: use machine data to predict when equipment will require maintainance. - [NEW 🚀] Amazon Monitron: End-to-end equipment monitoring system to enable predictive maintanance. - [NEW 🚀] Amazon Lookout for Equipment - Sends sensor data to AWS to build a ML model - Pulls data from machine operations systems, such as OSISoft - Learns normal patterns and creates a model - Uses real-time data to identify early warning signs that could lead to machine failures - 2. Computer Vision: improve processes, identify bottlenecks, and detect anomalies. - [NEW 🚀] Amazon Panorama Appliance - Add computer vision to your existing onsite cameras with AWS Panorama Appliance - Plug in appliance, connects to network, and starts to identify video streams from existing cameras - Pre-built computer vision models in manufacturing, retail, construction, and other industried - Can also build models in SageMaker and deploy to Panorama - Integrates with AWS IoT services, including SiteWise, to send data for broader analysis - [NEW 🚀] Amazon Panorama SDK - Enables hardware vendors to build new cameras that run more meaningful computer vision models at the edge - Provides camera manufacturers with an SDK and APIs to create cameras to run CV models at the edge - Chips designed for CV and deep learning from Nvidia and Ambarella - Panorama-compatible cameras work out of the box with AWS ML services - Build and train models in SageMaker and deploy to cameras with a single click

5. Hybrid


Guest: Zach Blitz, Head of Infrastructure, Riot Games

  • “With Outposts, AWS gave us a unique solution to ensure a level playing field for our players and streamlined our deployments using the same tools and APIs on-premises and in the cloud.”

What is hybrid?

若用二分法來看 hybrid 非常容易進到誤區,而無法看到 hybrid 整局全貌。

  • Early confusion: A binary solution and new software stack?
    • On-premises
    • Cloud
  • So, really, what is hybrid?

超想要玩小台的 AWS Outposts!看到簡報我已經想到一些可能的產品規劃了,來看看明後年有沒有機會玩到 :)

  • When trying to move workloads from on-premises to cloud - VMware Cloud on AWS - Customers reduced IT infra costs by 40% - Reduced total cost of operations by 43% - Estimated 479% ROI over five years
  • How do I gest AWS on-premises as I'm transitioning and for workloads that must remain on-premises?
    • AWS Outposts
      • Run AWS infra and services on-premises
      • AWS servers with AWS compute, storage, database, and analytics services
      • Fully managed and supported by AWS
      • Same hardware that AWS runs in its data centers
      • Same APIs, same control plane, same tools and same functionality. - [NEW 🚀] AWS Outposts in two new sizes. (1U, 2U) (Coming in 2021) - AWS Outposts for any location - Two sizes to fit locations with limited space - Provide 64 vCPU, 128GiB memory, 4 TB local NVMe storage - Fully managed by AWS

即將收尾的最後,快速介紹了距離全美大城市更近距離的「AWS Local Zones」、適合嚴苛環境的「AWS Snow Family」、5G 應用接地氣的「AWS Wavelength」,大抵上就跟無線通訊類似思維,藉由不同的距離(地理位置)切割出對應的技術架構與產品線,來滿足客戶的需求。

  • On-premises in major metros for your most latency-sensitive apps?
    • AWS Local Zones
      • Extension of an AWS Region
      • Deliver single-digit millisecond latency to local end users
      • Launched in Los Angeles in 2019 - [NEW 🚀] New in Boston, Houston, Miami (Preview) - 12 more coming in 2021! Atlanta, Chicago, Dallas, Denver, Kansas City, Las Vegas, Minneapolis, New York, Philadelphia, Phoenix, Portland, and Seattle. - 手癢換成該城市的主要機場代號: ATL, ORD, DFW, DEN, MCI, LAS, MSP, JFK, PHL, PHX, PDX, SEA。(都是重要的航空轉運基地,不知道是不是有更深層的原因先選定了這些城市。)

適合嚴苛環境且不異取得連線的場域,可以部署小到一個手提箱,大到一整台貨櫃車的 Snow 系列。

  • On-premises in poorly connected and rugged locations?
    • AWS Snow Family

那 5G 場域何解?

  • On-premises at 5G Edges?
    • AWS Wavelength
      • Build applications that serve mobile end users with ultra-low latency
      • Same AWS APIs, tools, and funcionality
      • Deploy 5G applications globally - Verizon in 8 US cities - KDDI in Tokyo and SK Telecom in Daejeon (Coming soon) - Vodafone in London (2021)

Andy 心中的 hybrid 從發生場域的近處到遠處,好比無線通訊由近至遠的 NFC、Bluetooth、Wi-Fi、Mobile LTE 等的場域分群、技術分群。

  • So, what’s hybrid? - On-premises at any facility - 5G networks - Major metros - Rugged edge

嗯,這張拍得有點俏皮,也是個 happy ending 囉 :p

AWS 團隊成員們大家在家裡頭用 Amazon Chime 連進來一起參與這場 Keynote 直播,側邊的這個副螢幕使用得相當恰當!

最後我就用這場 Keynote 的特別來賓之一,Snap Inc. 的 Jerry 分享的這句話來做為這篇記錄的結尾吧:「You might think that there isn't much left to reinvent when it comes to compute, but the innovations just keep coming.

這次發佈的新產品或新功能中,有哪些是你想拿來試試看、打造新產品或新玩具的嗎?歡迎來 AWS User Group Taiwan 跟大家一起聊聊吧!


  • AWS Cloud Products Study Notes Series will separate into couple tables. Each table has one perspective. In the end, I will summarize all in one.
    • Main Product List: This is the main entrance of this series, and you can start looking for various observation angles or new products from here.
    • URL Path List: Official website URL path analysis official website URL path to observe their classification and affiliation.
    • SLA List: SLA (Service Level Agreement) and availability.

Loading comments…