Last week I attended the Tech-Conference of Amazon Web Services – AWS re:Invent 2017 in Las Vegas. It lasted five days, a period of time that is not always easy to take off from your daily work. Following are the most important pieces of content from my perspective in 7-10 min for reading.
* 10 Seconds Management Spoiler *
Serverless, Machine Learning, the Machine Learning Camera DeepLens, Alexa for Business and Kube as a managed service are the main highlights of this year’s re:Invent. By extending and making existing and established services such as EC2, S3, Glacier or DynamoDB more flexible, AWS helps customers to map many requirements directly in the managed service and reduce the need for workaround implementations. It will be fascinating and at times frightening, what will be possible in the future due to the combination of these powerful services.
Amazon Web Services demonstrated at its own superlative Tech Conference – AWS re:Invent – in Las Vegas more than impressively who the global leader in the cloud business is. Furthermore, AWS has also shown that they are not resting on their laurels, but continue to set and drive the pace of global change in IT through continuous innovation and technological competence. In my opinion, there is another fact which ensures that this will remain so for some time to come, namely that the image of AWS has been carefully built up and maintained. AWS does everything to support customers and users. Customers get a huge set of components and services that fit together seamlessly, allowing them to concentrate on their core business tasks much better. And for the worldwide developer community, AWS is also extremely attractive: they are courted, supported and valued, and this pays off in order to continue to lead the market.
Back to the conference. More than 43,000 participants have registered and participated in more than 1,300 breakout sessions, workshops and demos over five days at six hotels along the Las Vegas Boulevard. The keynotes of Andy Jessy (CEO AWS) and Werner Vogels (CTO Amazon and AWS Mastermind) together lasted over six hours and included more than 25 major announcements of existing services and completely new products. Important innovations can be found in several areas, in the following please check my list of this year’s re:Invent announcements. However, there were so many announcements and in detail so many product improvements that only the most important announcements are described here from my personal perspective.
1) Infrastructure + Storage
The highlight at the infrastructure level is certainly the integration of Kubernetes as a Managed Service (EKS – Elastic Container Service for Kubernetes), which is expected by many people. AWS fully relies on compatibility with the official Kubernetes releases but integrates Kube with AWS‘ own network capabilities (VPC) and access controls (IAM). AWS Fargate for ECS and in the future also for EKS makes it possible to roll out and operate containerized applications without any administration effort.
S3 Select and Glacier Select allows to query subsets of existing S3 objects directly, which improves performance and reduces own workaround solutions. In the case of Glacier Select, archived and thus actually dead data can also be queried within minutes, and this data can then also be used for analysis purposes, thus experiencing a completely new significance.
EC2 Elastic GPU: GPU services vary widely in demand, and the use of EC2 GPU instances often requires provisioning of appropriate instance sizes without the need for GPU performance on a permanent basis. With Amazon EC2 Elastic GPU, GPU performance can be configured independently of the used EC2 instance and flexibly according to actual needs, i.e. if the need for GPU performance is high, scaling is increased, if the need for GPU performance decreases, GPU performance is reduced again.
2) Lambda and Serverless
.NET Core and Go as programming language for Lambda functions will be supported in the future and the memory limit for Lambda functions will be increased to three GB. With the AWS Serverless Application Repository, third-party providers will also be able to offer their serverless applications to other customers, in principle a form of marketplace for serverless applications. Overall, AWS also aligns other services for serverless applications and supports the development of serverless applications.
3) Cloud9 IDE
AWS has now its own IDE, which runs completely in the browser. This significantly improves the development of serverless functions for development, deployment and testing, including local testing of functions and Cloud9 IDE is closing the gap in the development of serverless applications. Cloud9 also supports the pairing of developers very strongly, i.e. they can work together, remotely or locally, on the same code. Other collaboration methods are supported, for example release workflows in the team or chat functions.
More extensive enhancements to existing database services are available for AWS Aurora database (the fastest growing service at AWS) and DynamoDB. Aurora for Serverless supports when workloads fluctuate or are difficult to predict and scales flexibly with actual demand. With Aurora Multimaster, worldwide zero downtime scenarios can be supported.
With DynamoDB Global Tables and DynamoDB On-Demand Backup, data is automatically replicated between AWS regions and can be backed up in a standardized way. With DynamoDB, AWS closes the gaps you expect from a fully managed service without having to implement your own workaround solutions.
Amazon Neptune as a completely new database service completes the NoSQL portfolio of AWS with a graph database. Amazon Neptune’s field of application is the handling of highly networked information, Neptune supports the most common frameworks TinkerPop and SPARQL for describing and querying graphs.
AWS now offers managed solutions in the entire NoSQL database context. DynamoDB as a database for key value and documents, Elasticache for in-memory and finally Neptune as graph database.
5) Machine Learning
This year’s and actual star among all the announcements is Machine Learning. Here AWS invests with power and you notice that they don’t want to leave the market to Google and Microsoft. AWS Sagemaker enables developers to configure machine learning algorithms and use them for their own applications without having to have too much background knowledge in the algorithms. Amazon is thus accommodating the developers. Sagemaker can also be used for training models based on Apache MXNet and Google’s Tensorflow.
AWS DeepLens is a camera optimized for machine learning scenarios and supports business models where image analysis and real-time response are important, Big Brother for everyone and every business model. In my opinion, Amazon’s next hardware hit and finally the proof that AWS and Amazon are leading in innovation. It will not be long before Google and Microsoft follow suit with their own hardware.
The services AWS Transcribe, AWS Translate and AWS Comprehend are provided for voice analysis and processing. Kinesis Video Streams and Amazon Rekognition provide the basis for custom solutions for real-time video processing.
6) Alexa and Voice User Interfaces
With the Echo devices and the opening of Alexa-Skill development to the developer community, Amazon has set the standard for voice interfaces, especially in the consumer market. Alexa for Business expands its portfolio now in the direction of business. Not only can it be used to develop and cover new applications, Amazon has demonstrated how to increase the efficiency of meetings through voice control, but it also simplifies the administration of a large number of voice-controlled devices in the company. It will be exciting to see how the Alexa community reacts to this, there will probably be a lot of new and innovative solutions.
7) AWS IoT
Six new service announcements alone were announced at re:Invent for AWS IoT. With IoT Device Management and IoT Device Defender, AWS simplifies the management of a large number of IoT devices and the creation and maintenance of corresponding security policies. IoT Analytics supports the special analytics requirements on IoT devices, Greengrass ML Inference brings Machine Learning capabilities directly to the IoT device and thus reduces transport and preprocessing of device data in the AWS cloud. With Free RTOS, AWS finally introduces its own device operating system to the market in order to meet the special requirements of preprocessing data directly on the devices.
Lot of fun was participating in a workshop on the preview of Amazon Sumerian, a 3D development environment for VR and AR that runs directly in the browser and is relatively easy to use compared to existing 3D engines. Sumerian offers a high level of integration with Amazon products such as Echo devices and AWS services and it can play a major role in improving the user experience for own business models. Currently Sumerian can only be tested in a pre-release version, but should be kept in mind in the medium term, especially for sales and marketing activities.
In addition to the service announcements, there is also a standardization in software development and cloud infrastructure that not only affects AWS, but in which AWS, like other vendors, is guided and supported by tried and tested procedures and best practices. Support for pairing in software development in the Cloud9 is an indication of this, and the use of standard frameworks such as Kubernetes or Google’s Tensorflow or Go for Serverless functions clearly underscore this.
However, the most important feature that runs through all services at AWS, is the importance of IT security for data and service endpoints, which is non-negotiable and not really visible. Luckily, IT security is positioned and implemented very prominently at AWS.