Voice As A Control SystemOne of the themes Vogels spoke about was IoT and allowing whole environments to become accessible. Every device has the ability to become an input or output device, but with so many out there, it's good to consider how we interact with all of them and their systems. Vogels believes that digital interfaces of the future will be human-centric, and the things that we as humans use to communicate will become the inputs to systems. The first of these will be the voice as it's the most natural and easiest interaction. Once you can use your voice to control systems, Vogel suggested people won't look back, from surgeons operating theatre equipment, to simply controlling the lighting or heating in your house, it will unlock digital systems for everyone. To demonstrate this point, Vogels talked about the International Rice Research Institute who provide rice farmers advice on how much and which fertiliser to put on their crops based on their years of research. Consumption of this information was very low until they invested in a voice interface. Farmers can call, select from one of 27 dialects, and provide information on their land and crop conditions. They then use voice recognition and machine learning to read back to the farmer which fertilizer they need. This was building up to the announcement of Alexa Business, a service that 'makes it easier for you to introduce Alexa to your organization, providing the tools you need to set-up and manage Alexa enabled devices, enroll users, and assign skills at scale'
Ensure You Are Well ArchitectedThe next theme of the keynote was architecture. Typically, systems have three planes: Admin, Control, and Data. (Vogels suggested architecture that extensive was difficult to visualise on marketing slides!) The AWS Well Architected Framework was launched two years ago and has grown from a single document to five pillars across five documents with two 'lenses'. It guides the user on how to architect for specific use cases, (currently HPC and Serverless). The framework is included in AWS certifications and AWS regularly run boot camps and 'Well Architected Reviews' for its customers.
Everything Will Fail. All The TimeAvailability, reliability, and resilience were discussed, from the basics, (hard dependencies reduce availability, redundant dependencies increase availability) to the best practices of Distributed Systems, through to deployment automation and testing. Nora Jones (Netflix), gave the example of using Chaos Engineering and how they do this at Netflix. Vogels highlighted that available systems cost more and therefore it becomes a business decision whether to easily run something in a single availability zone, but only achieve 99% of uptime. If you want to increase this you need to distribute your services across multiple availability zones or even regions. DynamoDB Global Tables, for example, help you to do this, becoming the ultimate tool in reliability design. Although this has little to do with AWS (and more to do with decisions made within organisations), AWS can make this much easier for you. This brings us nicely onto the final part of the keynote - letting AWS do the 'heavy lifting' through its managed services. Galls Law says, "A complex system that worked is invariably found to have evolved from a simple system that worked". AWS allows you to keep your systems simple by providing nimble services which you can assemble to build what you need. If you run your own RDBMS, you have to take care of the control and data planes. If you run on AWS, AWS manages the control plane. AWS Managed Services are designed for AWS to control the complex and hard to manage moving parts. making it simpler for you. This was demonstrated by Abby Fuller speaking about containers on AWS, and how Amazon Fargate can help you to make your environment much more simple. AWS will continue to release managed services over the next year.
ServerlessServerless was something that couldn't possibly be missed out of this keynote, with it being the ultimate AWS Managed Service. There is no server management, has flexible scaling, high availability, and no idle capacity. Here are the final (Lambda) product announcements
- API Gateway VPC integration
- Concurrency Controls
- 3GB Memory Support &
- .NetCoe 2.0 & GO language support
Get in Touch.
Let’s discuss how we can help with your cloud journey. Our experts are standing by to talk about your migration, modernisation, development and skills challenges.