The best of Re: Invent 2017

The best of Re: Invent 2017

Between november 27 and december 1st, we participated in Re: INVENT 2017, the annual event of Amazon Web Services, which was held in Las Vegas. From the vision of Xavier Verdino, in this post we share the most prominent announcements about new features and solutions of managed cloud services.

By Xavier Verdino

In a tough race in which AWS and Azure take advantage, cloud world moves by leaps and bounds and none of the key players wants to lose ground.

As in other years, the number of announcements of new features and services is very wide and it is impossible to go into detail about each of them. The idea in this article is to focus on the services that we understand most interesting from an IT / Devops point of view, leaving aside the announcements oriented to machine learning, cognitive skills, artificial intelligence, IoT, etc.

Within this framework some of the new services that were presented and we found most interesting are the following:

  • Amazon ECS for Kubernetes (EKS)
  • AWS Fargate
  • Amazon EC2 Bare Metal Instances
  • Amazon Aurora Multi-Master
  • Amazon DynamoDB Global Tables
  • AWS Serverless Application Repository



Amazon ECS for Kubernetes (EKS)

As defined by AWS, the Amazon Elastic Container for Kubernetes (Amazon EKS) is a managed service that facilitates the execution of Kubernetes in AWS without the need to install and operate your own Kubernetes cluster.

For all of us who have worked managing a traditional kubernetes cluster, it is common to find challenges such as administration and assurance of availability of master nodes, management of shared storage, patching of nodes, correct sizing of minions, etc.

Well … EKS comes to solve these problems, having three master nodes geographically distributed in three areas of availability and providing a service to detect problems in these nodes and automatic replacement. Thus, the management of these nodes ceases to be a headache.

It is important to clarify that EKS is 100% compatible with all applications that can currently run in a traditional K8s cluster.

Amazon ECS for Kubernetes (EKS)

AWS Fargate

Being able to deploy containers without the need to manage a cluster is something that sounds very good …. Well AWS Fargate is a service that allows you to operate containers without having to worry about the underlying infrastructure.

Fargate enters the scene to compete with Azure’s Conteiner Instance and Container Engine of GoogleCloud and intends to provide a service where the administrator should only put the energies in the container itself and forget cluster, vms, HA, etc.

One definition that I found very interesting was the one that Randall Hunt (Technical Evangelist at Amazon Web Services) makes in one of his publications “To put it in simple terms, Fargate is like EC2, but instead of giving it a virtual machine, it gets a container. It is a technology that allows you to use containers as a fundamental computer primitive without having to manage the instances”.

Going a little more in detail, Fargate can be used in two modes:

  • Fargate launch type: with this launch type the only thing that the administrator should worry about is to package their application in containers, then define the resource policies for this container (cpu and memory) and security policies, with this application is ready to be launched.


  • EC2 launch type: using fargate with the EC2 lauch type, you can choose to manage an ECS or EKS cluster and fargate happens to be an advanced orchestrator for these clusters. In this case, the operation and administration of the cluster is once again the responsibility of the administrator and not of AWS, which is responsible for provisioning, patching and scaling of server clusters. It allows much more flexibility when accessing the nodes but of course it generates an over work compared to the fargate lauch type.

AWS Fargate


Amazon EC2 Bare Metal Instances

AWS EC2 Bare Metal provides a traditional EC2 instance but with direct access to host resources. These types of instances are ideal for running applications that need access to hw direct features without going through the virtualization layer.

This service is still in preview so we could not prove it but it promises to be very useful for particular situations.

Amazon Aurora Multi-Master

In my opinion, one of the great announcements and a really new functionality in AWS that seems very useful.

Amazon Aurora is a database engine developed and offered as a service by AWS, at the moment it is 100% compatible with MySQL and PostgreSQL. A few months ago, AWS announced that Aurora allowed automatic scaling in replicated reads, meaning that you could automatically create read replicas on which the read queries were going to be balanced, which is very useful but under certain circumstances can be insufficient.

AWS announced that from now on Aurora can create multiple read and write instances in multiple Avaliability Zones. This allows applications to read and write data to multiple instances of databases within a cluster in a manner similar to existing read replicas.

This, according to AWS, “makes Amazon Aurora the first relational database service to scale both readings and writes across multiple data centers.”

Amazon DynamoDB Global Tables

AWS DynamoDB Global Tables is a new functionality that is integrated with the already known DynamoDB to allow creating tables that are automatically replicated between two AWS regions. Given this, from now on you can achieve fast read and write access that can scale automatically to meet the needs of the most demanding global applications.


AWS Serverless Application Repository

AWS Serverless Application Repository is a collection of applications published by developers and some companies. These applications are packaged using the AWS Server Model AWS Server Model (AWS SAM) and it allows to quickly deploy standard applications from code templates. It seeks to generate a repository from which the base can be built for any serverless application. It is also a service that is still in preview but it is to be assumed that the success of it depends on the contribution of the community.


Other services announced

It is impossible to go into detail for each of the new services and functionalities so I leave the list with the link to each one so that they can delve into the ones that make the most sense.



  • Amazon Kinesis Video Streams (


Application Integration

  • Amazon MQ (



  • Amazon Sumerian (


Business Productivity

  • Alexa for Business (



  • Amazon EC2 M5 Instances (
  • Amazon EC2 H1 Instances (
  • New Deployment Options for AWS Lambda Functions (
  • Hibernation for Spot Instances (
  • New Spot Pricing Model and Spot on RunInstances (
  • Array Jobs for AWS Batch (
  • Amazon Lightsail Load Balancer (
  • Launch Templates for Amazon EC2 Instances (
  • Spread Placement Groups for Amazon EC2 (
  • Amazon EC2 T2 Unlimited (
  • Amazon Time Sync Service (
  • AWS Lambda Doubles Maximum Memory Capacity for Lambda Functions (
  • AWS Lambda Per Function Concurrency/Throttle (
  • AWS Lambda Introduces Enhanced Console Experience (
  • Migrate Hyper-V VMs to AWS with AWS Server Migration Service (



  • Amazon Neptune (
  • Amazon DynamoDB On-Demand Backup (
  • Amazon Aurora Serverless (


Developer Tools

  • AWS Cloud9 (
  • Support for Serverless Deployments in AWS CodeDeploy (



  • AWS IoT Device Management (
  • AWS IoT Analytics (
  • AWS IoT 1-Click (
  • AWS IoT Device Defender (
  • Amazon FreeRTOS (
  • AWS Greengrass ML Inference (
  • AWS Greengrass (
  • AWS IoT Core (