AWS Re:Invent 2018 - An Overview

The jet lag is wearing off, the cold weather of England is back and the bright lights of Las Vegas are way off in the distance. So now seems like a good time for a brief recap of my personal highlights from AWS Re:Invent 2018.

Highlights

DynamoDB

DynamoDB had some really exciting features making their public release at Re:Invent. Support for Transactions is a big win for engineers and allows you to write all-or-nothing changes in a much simpler way than before. On-Demand is the other DynamoDB announcement that grabbed my attention last week; I’ve always found the capacity provisioning for a DynamoDB table difficult, and often ended up over-provisioning as a result. Now DynamoDB allows you to create a table without having to plan your capacity upfront, allowing you to understand your traffic patterns rather than guessing at them.

CloudWatch Insights

The CloudWatch team had a relatively quiet week at Re:Invent but I was glad to see the public release of CloudWatch Insights .

As you can see from the above diagram, CloudWatch Insights brings fast, interactive log analysis natively to CloudWatch. I’ve used tools such as SumoLogic and Splunk in the past and I’m excited to see how Insights compares.

Serverless

Just as Werner teased before his Keynote, Serverless was a big focus point and as such was accompanied with some really cool announcements.

Firecracker

Whilst most of the Serverless features were (naturally) aimed at users of AWS Lambda, one of the more intriguing releases of the entire week was Firecracker . Firecracker is an open-source virtualisation technology that allows you to create and manage containers. It’s used to power both AWS Lambda and AWS Fargate so it’s great to see AWS opening this up to the open-source community.

ALB ability to invoke Lambda functions

Personally, this is definitely one of the most exciting announcements from Re:Invent - invoking a Lambda function using an Application Load Balancer . Yes, invoking a Lambda over HTTP(S) was possible previously using API Gateway but the ability to have your Lambda functions and AutoScaling groups fronted by the same load balancer is a big win.

Custom Lambda runtimes

Another release which I’m sure made the weeks of many of the attendees last week. Custom Lambda runtimes allow developers to maintain their own runtimes, allowing you to use custom binaries, programming languages and versions without having to rely on AWS to create and manage them. As a engineer who primarily writes Node.js applications I’m really excited at the possibility of having Lambda functions that can take advantage of newer language features without having to wait for an official runtime to be released.

Lambda Layers

Released at the same time as custom runtimes was the concept of Lambda Layers . Layers are an artifact which can be referenced by any many or as little Lambda functions as you like. These artifacts are uploaded centrally and individual Lambda functions can then be configured to reference these layers which are then available at runtime.

Final thoughts

AWS Re:Invent was a great week, organised exceptionally well (sorting out a conference attended by more than 50,000 is no small feat!) with some fantastic features released as well. I’ll be publishing another blog post soon focusing with some tips for attending Re:Invent, so keep your eyes peeled!