I had the luck (and I say this considering the number of talented engineers within Shine) of attending AWS re:Invent in Las Vegas recently.
For those who don’t know, re:Invent is an annual conference held by AWS. It’s a chance for customers, vendors and AWS staff to come together and learn more about the platform and get the latest information on new AWS releases.
For a first time attendee, it can be quite an overwhelming experience! Nearly 50,000 attendees spread amongst seven different venues across South Las Vegas Boulevard.
Below I’m going to take a look at some of the major announcements of the event and some of the talks I found most interesting.
Firecracker – Possibly one of the most interesting announcements at the event, and it’s not even a service you can directly buy. Firecracker is a MicroVM technology created by AWS that they are currently using to power AWS Lambda and Fargate. Unlike a traditional heavyweight full virtual machine, Firecracker VMs take advantage of modern hardware virtualisation to create VMs that can launch quickly (~ 125ms) with low overhead (5MB memory). This may not seem like a particularly glamorous announcement, but it could be a game changer in the world of virtual machines and containers. I’d recommend checking out this video from the event for deeper insights.
A1 Instances – AWS custom-built Arm CPUs, what’s not to like? Immediate significant cost savings are available with these. It’s going to to be an interesting space to watch as AWS expands into more custom-built hardware (look into the Nitro hypervisor for custom hardware they are already running).
Custom Lambda runtimes – Cobol on Lambda, why not? After releasing C++ and Rust runtimes for Lambda, AWS opened it up further, and now allows you to build custom runtimes to run pretty much any language you want. As part of this, they also introduced lambda layers, so you can share common functions across more than one lambda.
DynamoDB – On-Demand – This one was huge. For those who don’t know, in DynamoDB you previously had to manage read and write capacity (i.e. the number of read/writes you can do in a space of time). If you breached this you would be throttled. AWS did provide auto-scaling, but you still had to tune this to get the desired response. Now you can enable On-Demand and AWS takes care of the rest (one caveat: keep an eye on your bill, as you are charged by read/write capacity).
Transit Gateway – Transit VPCs are a common practice in AWS, used for wiring together multiple VPCs and customer networks. The newly announced Transit Gateway feature replaces a lot of the more granular steps that you would have had to implement in the past to do this wiring.
Least surprising announcement
Outposts – It’s been a long time coming, but AWS has finally acknowledged that 100% public cloud may not be the only way. With this announcement, you will be able to buy AWS-managed servers for your data centre. You will be able to manage this infrastructure with the same native AWS APIs that you use to manage your public cloud infrastructure.
Most surprising announcement
Ground Station – Satellites as a Service. I’m pretty sure no-one saw this coming. AWS tends to build the things they need first and then sell them to the public. It will be interesting to see what they have up their sleeve with this one.
Below are some of my favourite sessions from the conference.
A Serverless Journey: Under the Hood of AWS Lambda – One of the best sessions I attended at the conference. If you’ve ever wondered about how AWS orchestrate trillions of Lambda executions every day, then this talk will give some insights into the internals of the Lambda platform.
Advanced Design Patterns for DynamoDB – If you’ve ever wanted to know more about the intricacies and complexity of modelling data effectively when using a NoSQL database then watch this talk!
Another Day in the Life of a Cloud Network Engineer at Netflix – Donavan Fritz and Joel Kodama from Netflix run us through some of the significant engineering challenges of operating a global network at Netflix scale. Most engineers will never face these challenges, but interesting stuff nonetheless.
Container Power Hour with Jess, Clare, and Abby – Jess Frazelle (Github), Clare Liguori (AWS) and Abby Fuller (AWS) run through some of the deployment options for containers on AWS Fargate. This includes a demo of using the new Github Actions to deploy a container to Fargate.
Inside AWS: Technology Choices for Modern Applications – A great talk from Tim Bray. Tim walks us through some of the modern application patterns and practices of successful customers and internal teams at AWS.
Here are some tips I think might be helpful for first-time attendees.
- If you’re familiar with AWS already then push yourself to go to sessions that might be the edge of your comfort zone. I probably spent the first two days going to sessions about technologies I was already familiar with, thinking it might be relevant for my current role.
- After every session, presenters will hang around and answer questions outside of the venue. Take advantage of this to ask questions. It’s not often you have access to principal engineers of AWS.
- You’ll find some of the biggest tech companies in the world have booths in the Expo Hall. Again, if you have questions about technologies like Gitlab, Github, and Docker, you will find engineers at their booths that you can speak to.
- Mix up your session attendance with chalk talks. Chalk talks offer a smaller, more personal experience. You can not only speak to AWS presenters at these talks but also with your fellow professionals.
- Attend the after-hours social events! It’s an opportunity to meet and talk tech with people from many diverse backgrounds.
So after a couple of marathon flights and a crazy week, it’s back to reality. All in all, it was a fantastic experience, with an opportunity to gain insights into the leading edge of cloud computing in 2018.
I will look to follow up this blog with a couple of hands-on articles related to some of the new technology announcements at re:Invent.