What's New from AWS Re:Invent 2018?

December 03, 2018

Down arrow black


Head of Development

AWS re:Invent 2018 has wrapped up and like always, the year’s most anticipated and celebrated 4-day cloud conference held at Las Vegas unleashed a barrage of exciting product and service announcements. They reaffirm AWS’s intention to remain at the forefront of serverless leadership - and how! Roughly 240 sessions were in some way about serverless topics.

It wouldn’t be humanly possible to list all my favourite announcements but here’s a few that I’m most excited about:

1. Cloudwatch Insights

Logging has been one of the long-standing pains of working in a serverless environment, and an expensive one to remedy. It was in fact one of the top voted-for topics at our community led SLS.zone meetup (and if you haven’t already check out our article on logs here). I ‘m excited to try out this cloudwatch update and see if this helps.

2. AWS Lambda Layers

At its basic level, AWS Lambda layers permits code sharing between serverless functions. One use case is the pre-packaging of application dependencies (library code as well as binaries) to reduce the amount of code a developer has to manage, duplicate and deploy. This new feature is really a huge win for those of us who run our functions with hefty binaries like FFMPEG for multimedia or Puppeteer for browser automation. It’s now possible to upload such software once and deploy functions quicker because they’re smaller. A point to note is that AWS Lambda layers are merged into your functions at create or update time, and account towards deployment package size limits.

3. AWS Lambda Runtime API

The real power of layers though is the introduction of the runtime API and the ability to run custom runtimes. To put it simply, developers are now granted full control over how they want their execution environments to run for their serverless applications. Use cases, in addition to the pre-packaging you’d now expect with layers, could be using non-officially supported languages or language versions in AWS Lambda, abstracting away tracing and logging middleware from application level to runtime level or satisfying mandated IT security policy by using proprietary or bespoke runtimes/packages - the possibilities are really endless!

However, this is probably the reason why I wouldn’t hurry to migrate clients to custom runtimes anytime soon. Whilst the new capabilities are powerful, the danger would be introducing yet another (and a multitude more complicated) dependency to manage and maintain for the long term. We’re using managed services after all, this should be taken care of for us! For a majority of use cases, I believe AWS Lambda stock runtimes are perfectly fine.

4. AWS API Gateway Websockets

Unfortunately not much has been shared on APIG Websockets support - only that it exists and more information will be revealed soon.

An exciting year ahead for Serverless at Reason

Here at Reason, we’re excited to be building real world, production-ready serverless applications for our clients and promoting serverless architecture to our peers. With the enormous backing of AWS as shown in this year’s Re:Invent, we’re expecting an amazing year ahead in the serverless space.

If you haven’t already, check out sls.zone - it’s our collection of tutorials, best practices and open source projects created for the serverless community to help build serverless applications the right way. If you’re interested in working with serverless technology, take a look at our careers page - we’re hiring at all levels.

Tweet me at @jimle_uk