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](https://aws.amazon.com/blogs/aws/ec2-instance-type-update-t2-r4-f1-elastic-gpus-i3-c5/)— More options for compute
General Purpose — t2.xlarge, t2.2xlarge more CPU and compute
Memory Optimised — r4
Storage Optimised — I3
Compute Optimised — C5
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](https://aws.amazon.com/blogs/aws/amazon-lightsail-the-power-of-aws-the-simplicity-of-a-vps/) —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 — https://amazonlightsail.com/
Availability: Now
[Hardware acceleration](https://aws.amazon.com/blogs/aws/developer-preview-ec2-instances-f1-with-programmable-hardware/) — 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
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](https://aws.amazon.com/blogs/aws/amazon-athena-interactive-sql-queries-for-data-in-amazon-s3/) — 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](https://aws.amazon.com/blogs/aws/amazon-aurora-update-postgresql-compatibility/) — New database engine for Aurora
High performance
Low cost
Availability: Preview
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](https://aws.amazon.com/blogs/aws/amazon-lex-build-conversational-voice-text-interfaces/) — 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
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](https://aws.amazon.com/blogs/aws/aws-greengrass-ubiquitous-real-world-computing/) — 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](https://aws.amazon.com/blogs/aws/aws-snowball-edge-more-storage-local-endpoints-lambda-functions/) — Hybrid Device with Storage and Compute
100 TB Storage
S3 endpoint
Greengrass (Lambda) inside — equivalent of m4.4xl inside
Cluster Snowballs
Available: Now
[Snowmobile](https://aws.amazon.com/blogs/aws/aws-snowmobile-move-exabytes-of-data-to-the-cloud-in-weeks/) — Clustered Snowballs in a truck
100 Petabyte container
Connect to your datacenter via fiber
Quickly move large amounts of data
Available: Now
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.