Andy Jassy, AWS CEO, was the first speaker out of the two main Keynotes from Re:Invent 2017, the AWS massive showcase event where all major updates and new products are unveiled. As expected, the announcements made by AWS were impressive and left us with the intention of getting to try them as soon as they become available on the Amazon Web Services Management Console.
Jassy´s keynote started with a clear message: “everything is everything”, meaning that AWS wants to bring everything to builders so that they can focus on being creative and give their best creating new products.
The AWS first main keynote 2017 had all the expected components, from live music (Foo Fighters or George Michael) to star guests such as Expedia’s CEO Mark Okerstrom or Michelle Mc-Kenna Doyle, NFL’s VP and AWS Machine Learning Doctor Matt Wood performing live demonstrations. In between the show, a list of new products, improvements and computing cloud innovations were presented and received with huge round of applauses.
Jassy classified the computing resources in three types:
The most famous AWS component, EC2 is already a consolidated item which can hold any kind of application and perform any task. Today is made of 13 families in total, the last of which are the Bare Metal Instances.
AWS announced this year that elastic GPU can be attached to most of the instances in those families.
Here we find one of the big shots this year with the arrival of EKS, which can be explained as the Elastic Container Service (ECS) supporting natively Kubernetes. And if that is not enough for you, AWS launched as well a service which handles the cluster and its management: AWS Fargate. It can be expected now that containers will take the lead in application deployment, and a lot of customers will migrate happily to this platform, as it brings all the benefits of containers while making the cluster management easy with Kubernetes.
There has not been any special announcement regarding this matter during the keynote, but we have to highlight the huge improvements that SAM (Serverless Application Model) has achieved in order to build a core for serverless applications CI/CD process.
In the Database section, the big announcements were the improvements on Aurora and DynamoDB as well as a new innovative service, AWS Neptune. The expectation and positive welcome to those big news were clearly manifested at the auditorium.
Those changes makes us eager to get started right away, in part with the arrival of the Auorar’s multi-master feature, which will allow having various read/write instances in multiple AZs. This will enhance the application’s scalability. The multi-regional option will be implemented during 2018.
The second main announcement concerning databases was Aurora Serverless: a service that will respond to unpredictable or cyclical workloads by auto-scaling as needed, avoiding the supply or management of instances.
DynamoDB has also had large announcements: the long- awaited Backup and Restore mechanism that will enable periodical backup executions and restore in a simple way, avoiding performance impact. We will have to wait until 2018 for point in time restore although.
Another announcement was DynamoDB Global Tables: a fully-managed multi-master and multi-region database. Such news are a delight to all applications that will be deployed at a global scale.
Lastly, AWS Neptune adds a new database layer. Up until now, databases were differentiated as relational or nonrelational. With AWS Neptune we can now add a third level: the Graph Database. The main objective is to improve the mining of highly connected data.
One of the most common AWS S3 use is as storage for Data Lakes, defined as a massive storage of raw data. In order to exploit the data, Amazon Web Services offers different options such as Amazon Athena, Amazon EMR, Amazon Redshift, Amazon Kinesis, Amazon QuickSight or AWS Glue, and now they have released a new one: AWS S3 Select. This product will allow us to make queries using the standard SQL language to filter the desired S3 subset without compromising the performance, even against petabytes of data.
In line to this product, it has also been announced Glacier Select, that enables to make queries over data stored on Glacier.
Such discipline is nowadays one of the strongest topics and all technological companies are investing and innovating on it. As Andy Jassy mentioned, they have been developing around Machine Learning for many years now, and the updates presented at this Re:Invent edition
aim at making this technology easier to use and implement for companies, since this is the big handicap they have complained about.
To accomplish this objective, Amazon Web Services has launched SageMaker, a service which will guide us through all the phases of Machine Learning: first you can start with a prebuilt algorithm or bring your own and run it on a list of different frameworks; next there is an excellent tool to help you train your model with just one-click and also you can tune the parameters easily; and finally it will be a great help in hosting and deployment automation.
Also, as we could see in a live demonstration during the event performed by Dr. Matt Wood, DeepLens will be soon in the market as the first HD camera with native support for Machine Learning.
And there is more, Amazon Web Services has gone further with Rekognition, adding the skill to process and analyse real-time batch videos: Rekognition Video will be able to detect objects, scenes and activities, track people, recognise celebrities and more. To help this service receive its input data, Amazon Kinesis Video has been launched too. By the combination of these two services, customers will be able to process real-time data from connected devices.
Voice and text services have also great news to announce: Transcribe, Translate and Comprehend can have a great impact on how new applications will interact with users. Transcribe will perform speech recognition, it will support multiple speakers and languages. Translate is about real-time translation with automatic language recognition. And Comprehend will be able to discover valuable insights from text documents, such as entities, key phrases, languages and feelings among others; it will also be able to classify those documents in categories.
We cannot stop thinking of the great number of applications that could be enhanced by using any of these services, and also our internal applications could be improved in order to help our clients and colleagues become more productive.
Internet of Things (IoT)
Last but not least, we got to hear about the next phase of devices connected to the Internet. Amazon Web Services announced a quite complete list of services which will help us with the device management:
- AWS IoT 1-click: one click creation of the Lambda trigger for any device IoT.
- AWS IoT Device Management: large scale self provisioning and remote management of devices.
- AWS IoT Device Defender: manage the security within devices. We are looking forward to testing this tool since security is one of the biggest concerns of our clients.
- AWS IoT Analytics: perform ad-hoc queries and cook the devices data so that they can be processed with Machine Learning services.
- Amazon FreeRTOS: operating system based on FreeRTOS for microcontrollers. Gather and securely send the data to AWS from the device will be its main purpose, as well as manage the keys to connect to AWS or to other services such as GreenGrass.
- AWS GreenGrass ML Inference: will bring the capacity to run Machine Learning on Edge.
To sum it up, this first keynote left us with an array of new products, features and important improvements and as recent upgraded AWS Premier Partners, we cannot wait to get hands on to try them on and apply them to the benefit of our clients. We will publish shortly the second blogpost highlighting the main points we think were most relevant from the second keynote at the AWS re:Invent 2017 directed by the Amazon.com CTO, Werner Vogels.