Reporting From the Front of AWS re:Invent 2016

By: Daniel Berman

LAS VEGAS, Nevada — AWS CEO Andy Jassy really enjoyed the keynote he gave yesterday at AWS re:Invent. Thirty-two thousand attendees watched him jab Oracle founder Larry Ellison while he spoke about Amazon providing its users with visibility and vision. Then — as a picture of Ellison showed up on the large screen on stage — Jassy mentioned that Amazon Web Services also gives the “ability to see through hand-waving and bombast.”

There was plenty of fanfare during the keynote, including an actual 45-foot semi-trailer riding onto the stage — called Snowmobile, the truck will transfer up to 100PB of data at a time to Amazon’s data center — but there were also some extremely interesting announcements about new services.

(For more reasons that Ellison might want to be less excitable, read our DevOps Pulse 2016 study. Among other industry findings, it shows that AWS rules the cloud — and Oracle is not even close.)

So, here’s a short recap of the major news that has come out of AWS re:Invent so far, just in case you happened to miss out on the live streaming of the event.

New Instances and Computation Services

Jassy announced a series of new instance types and services to complement and enhance Amazon’s computation offerings, including new T, R, I, and C family instances as well as a new instance family for FPGA instances. All new instances offer enhanced memory, throughput, and processing.

Also announced were Elastic GPUs for EC2, which will allow users to allow attaching GPUs to any of the instance types.

Other interesting news in the computation field included the introduction of a new Amazon service called Lightsail, a service that will easily create a VPS (virtual private server). Users will be able to get a VPS up and running in just a few steps while — behind the scenes — AWS will take care of launching the VM, attaching the SSD, and then configuring the IAM and security groups. (Of course, it will be interesting to see how companies that offer similar VPS hosting services such as DigitalOcean will react.)

Databases

The big news on Amazon’s Aurora database service is that it is now fully compatible with PostgreSQL. Jeff Barr covers the specifics in this blog post, but from a business perspective, this move will make AWS extremely appealing to enterprise companies that are dissuaded by Oracle and Microsoft’s prices.

Big Data Analysis

A long-awaited announcement in the realm of analysis is a new service called Athena.

This serverless service allows for the interactive and direct querying of S3 using standard SQL. There’s no cluster to handle and no management overhead — users can just query the data without worrying about the infrastructure.

Users will pay per query, but the query results get stored in an S3 bucket — meaning that people will need to pay additional charges for the S3 usage.

Machine Learning and AI

Judging from both Andy Jassy’s keynote and Werner Vogel’s post from yesterday, it seems that Amazon has taken to heart claims that they are lagging behind their competitors in the fields of machine understanding and artificial intelligence. Both made a point of highlighting Amazon’s legacy work and commitment in these fields before introducing the new services.

With more and more cloud providers offering machine learning tools and artificial intelligence as part of their standard enterprise cloud offerings as well as the emergence of a new and growing ecosystem around this field, I was expecting Amazon to make plenty of big announcements during the keynote. Indeed, there was plenty of news — but you can be the judge of how much of it was significant.

During the keynote, Jassy introduced a machine understanding toolkit for developers that consists of three complementing AI services:

Rekognition is a managed, deep-learning service for analyzing images and objects. These objects can be passed either in batch or in real-time to the service via SDK or API, both of which are integrated with S3 and Lambda.
Lex takes the technology that powers Alexa — Amazon’s voice-activated virtual assistant — and allows developers to use it to build apps using conversational interfaces with voice and text.
Polly is Amazon’s new TTS (text to speech) service that takes streams of text and produces a speech output.

“It’s all about having the right tools to build with,” Jassy told the audience at re:Invent. With these three services, developers have a solid toolkit to use when building the next generation of cloud applications.

IoT

Extending AWS onto Internet of Things devices is essentially what Greengrass, the last new service announced during the keynote, is all about.

Built on top of both AWS IoT and AWS Lambda, Greengrass will enable developers to write Lambda functions that run on the device. Greengrass will run these functions locally on the device but will also interact with AWS on the cloud for management purposes.