BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
AWS re:Invent, the annual AWS conference, has become one of the most recognized events in the public cloud space. Now in its fifth year, the sold-out AWS re:Invent 2016 promises two keynotes that will be jam-packed with new product announcements, updates to existing services such as Alexa and Lambda, price cuts and the occasional jab to other popular public cloud services on the market.
SearchAWS asked its experts to share their insights on what to expect from AWS re:Invent 2016 and which products will garner the most attention. Here's what they had to say.
What will be the big news coming out of this year's AWS conference?
KURT MARKO: AWS will make announcements across its service line; I expect incremental improvements and some price cuts in core infrastructure services such as Elastic Compute Cloud (EC2), storage and databases. I hope it addresses the cost of moving data out of AWS; it's a significant inhibitor to building hybrid clouds, including those built on VMware, following the services announced in October 2016.
Strategically, I expect AWS will make significant announcements around containers, serverless computing and machine learning. The AWS container service, Elastic Container Service (ECS), is a credible, Docker-compatible container service, but it's missing some important features compared to container orchestrator, Kubernetes. ECS needs support for a container and service registry, as well as better service discovery, configuration, load balancing and routing. These changes can come either through improvements to Elastic Load Balancer or a new, native container service such as Consul. I expect ECS will add some form of cluster resource optimization like Kubernetes automatic bin packing.
In machine learning, I expect AWS to introduce services built on its machine learning engine, making it easier for non-specialists to use the service. These might include prebuilt models for image and speech recognition, text/semantic analysis, and translation and predictive statistics. I also expect to see improvements for developers through support for third-party machine learning frameworks like Apache Singa, Caffe, TensorFlow -- iffy, given that Google developed this -- Theano and Torch.
Finally, I would like to see improvements and a roadmap for Workspaces, AWS' desktop as a service offering. The product seems to be languishing, but holds promise now that AWS has broadened its use of GPU compute instances and because organizations are more comfortable with tablets, Chromebooks and cheap PCs.
BRIAN TARBOX: Two years ago, the big deal was Docker. Last year was all about Lambda. This year the AWS conference might have a greater emphasis on Alexa, Echo and the internet of things (IoT). Alexa skills are fairly simple to write and provide a new way to bring technology to consumers. The real barrier to greater adoption of IoT at home is the price point; wirelessly controlled color-changing light bulbs cost $40 each. I'm hoping manufacturers use re:Invent to announce a dramatic price drop.
DAVID LINTHICUM: AWS will announce more around machine learning, containers and IoT, likely focusing on the next generation of the technologies. I also suspect that there will be some new big data technology announced that will further erode the dominance of the big enterprise data players, such as Oracle and IBM. The cloud provider will also release more tactical features and functions its users have asked for.
GEORGE LAWTON: The first generation of Alexa has been an experiment in creating an entirely new class of voice interfaces that blend simple hardware with complex cloud services. There are a number of technical challenges in enabling the voice interface so it's useful for mainstream consumers. This year, AWS will improve the service back end for more intelligent applications that add support for chatbot functionality, simplify integration for the technically illiterate and improve support for smart home devices.
AWS may also launch richer support for orchestrating the coordination of multiple voice interfaces into a common back end. This will make it easier for consumers to specify access and permissions across different household areas. AWS might also enable support for biometric audio authentication, which is required to regulate or restrict access to applications. Once these issues have been worked out in the home, Alexa will make more sense for enterprise uses cases. It is possible, but probably a little early, that AWS will add integration of enterprise identity and access management required for mainstream use in the enterprise. This is more likely to come in a beta form in late 2016.
The evolution of microservice architectures and continuous integration in the software development lifecycle is driving a trend toward smaller services with greater modularity. The Lambda platform is a candidate for weaving together simple code snippets with tight integration into design pattern around function as a service. Expect a number of important announcements for integrating Lambda support into other AWS applications for persistent storage, identity and access management, and analytics.
Additionally, the cloud provider likely will improve Lambda tooling to allow programmers to instantiate a collection of Lambda functions for common business processes and different types of integration. This could include better versioning, business logic components and improved IoT integration.
CHRIS MOYER: No surprise that we'll see more enterprise services. I expect to hear more details about the VMware integration, and possibly how that will affect Docker. I hope to see better Docker/ECS handling. As a matter of fact, I noticed today that my ECS services weren't working because an instance was still running, but the Docker agent was failing. AWS Auto Scaling didn't detect the issue and replace the server, because the server was still running.
It's silly to think we need to manage servers in the age of Docker and Lambda. I expect to see a lot more flexibility with both. Perhaps we'll see something that merges the two, allowing Lambda to run Docker Functions or having Docker as a first-class citizen in the AWS ecosystem -- without having to manage underlying ECS resources. I also expect to hear more about the new US-East Ohio region.