AWS re:Invent 2016 — Wednesday’s Keynote

A Summary of the Wednesday Keynote by Andy Jassy

After two big days of sessions and events at AWS re:Invent, Wednesday started with the first of two general keynotes. There was an air of excitement for the first round of announcements by AWS CEO, Andy Jassy. We saw a wide range of announcements for infrastructure and platform services.


AWS continued to improve their compute offerings this year with updates to their Infrastructure as a Service offerings. The new instance types and services provides tools for those who have found existing offerings do not quite fit their needs.

[EC2 Instance types](— More options for compute

Available: Now

**Elastic GPUs For EC2 — **Attach GPU to EC2 Instances

  • Similar to EBS you can attach GPU to instances

  • Options for 1,2,4,8 GiB of GPU

Availability: Preview

[Amazon Lightsail]( —Virtual Private Servers (VPS) made easy

  • Run virtual private servers without configuring a VPC

  • Choose image, Select size, Pick name and go

  • Simple price per month

  • Move Lightsail to AWS when requirements change

  • Get started —

Availability: Now

[Hardware acceleration]( — Programmable hardware

  • A new instance family — F1 instances

  • Develop Field Programmable Gate Arrays (FPGA) in AWS

  • Released a Hardware dev kit on GitHub (link doesn’t work yet?)

Availability: Preview

Data storage and services

We saw a couple of interesting new features for data storage in AWS. You cannot your run an access database on s3. But now you can query your S3 data with SQL.

[Amazon Athena]( — SQL queries for Data in S3

  • Ad-hoc query against s3 without clusters

  • Fully managed by AWS

  • Doesn’t replace Redshift or EMR — another option

Availability: Now

[Aurora for Postgres]( — New database engine for Aurora

  • High performance

  • Low cost

Availability: Preview

Amazon AI

The suite of Amazon AI services is an great addition to AWS. Image recognition and Natural Language Processing will allow more developers to create intelligent user experiences.

**Amazon Rekognition — **Image recognition service

  • Pass image to Rekognition in Batch or Realtime

  • Identifies Objects & Scenes — car, outside ect

  • Identifies Faces — gender, smiling, glaces, matching

  • Improve models over time

Availability: Now

Amazon Polly — Text to speech service

  • Convert text to an MP3

  • Fully managed, cached responses

  • 47 voices, 27 languages

Availability: Now

[Amazon Lex]( — What’s inside Amazon Alexa

  • Automatic Speech recognition (ASR) — speech to text

  • Natural Language understanding (NLU)

  • Processes text or audio — suitable for voice or chat bots

  • Triggers lambda to act upon requests

  • Can be used for multi-step conversations

  • Conversation models will improve over time

Available: Preview

Internet of Things (IoT)

I am personally very excited for updates to the IoT offerings on AWS. Project Greengrass is exactly what I need to finish my Serverless Garden project. Deploying software to devices and operating devices in an offline world are hard problems to solve. Project Greengrass will help solve this issue.

[Greengrass]( — Lambda compute on Devices

  • Embed lambda functions in devices

  • Runs locally and offline

  • Cache data locally

  • Manufacturers can build Greengrass into devices

  • Install Greengrass runtime

  • Deploy Lambda functions to devices

  • Facilitates device communication

Availability: Preview


[Snowball Edge]( — Hybrid Device with Storage and Compute

  • 100 TB Storage

  • S3 endpoint

  • Greengrass (Lambda) inside — equivalent of m4.4xl inside

  • Cluster Snowballs

Available: Now

[Snowmobile]( — Clustered Snowballs in a truck

  • 100 Petabyte container

  • Connect to your datacenter via fiber

  • Quickly move large amounts of data

Available: Now

What’s Next

We saw some great announcements today and we’re looking forward to more tomorrow. Andy hinted that tomorrow will be the day for Serverless announcements. I hope Werner Vogels might tick off some more items on my wish list tomorrow.

Thanks to my colleague Daniel Parker for the photos.