Serverless Superhero

Our serverless superhero this week is Ali Haydar, platform team lead at Fergus and AWS Community Builder. Ali is an active member of the serverless community where he does public speaking, blogging, and mentoring of others (whether he knows it or not). I always enjoy seeing Ali’s contributions online, be it on LinkedIn, Slack, or anywhere else the community meets up. Thank you for everything, Ali!

Tutorials

I tend to shy away from “getting started” posts in this newsletter, but I really like this one from Chibuike Nwachukwu. He goes through step-by-step how to build a RAG pipeline with Amazon Kendra and Bedrock for a chat app from scratch. He uses AWS Amplify for the frontend, hosting, and auth, and shows you how to create an API that will use your data as a knowledge base for a chatbot. This is the time of year for experiments, and this is a great tutorial to follow if you’re ready to get into GenAI.

With the emergence of LLMs and developers getting access to more and more massive data sets for real-time processing, the appeal of using EFS with Lambda to boost performance is going way up. Emma Moinat shared a post walking you through how to set it all up and explaining the important pieces. She even provides you with CDK code so you can get it deployed as quickly as possible.

We all heard Werner Vogels say it was time for Rust. So Ben Pyle has graciously taken us through how to use Rust with Fargate to build a web API. There’s so much in this article I can’t possibly summarize it all. He talks about the availability of useful crates in Rust, provides source code for communicating with DynamoDB in a container, and shows us how to define routes and models for an API. Then he goes on to talk about tuning load balancers, how to build a Rust-based Docker container, and shares the definitions of the tasks, clusters, and services. It’s so much good, practical information. Great one to end the year, Ben!

Interesting Reads

Lee Gilmore rounds out his incredible 2023 with a post on serverless idempotency with Lambda powertools. He gives a pragmatic example of when you need idempotency, shows the logical workflow of an idempotent request, and even shares a CDK construct that does it all for you. I love how easy Lee makes things to understand and implement. This is a thorough post, providing you with what feels like production-ready knowledge.

Simpler is better. That’s why when I read Matteo Depascale’s post on how to set up a call center on AWS, I started wondering a little bit. The blog is written clearly and posed in an easy-to-understand way, but it feels like a heavier build than necessary. The cost and amount of resources feel like one of those scenarios where you should use it if you’re committed to only using AWS. Matteo shows how easy it is to setup in Terraform, but from my personal experience, the right tool for the job here is Twilio. I highly value Matteo’s work on this post - it’s super interesting to see how AWS solves the call center problem.

Spotlight

Darryl Ruggles wrote a great blog post on how he created a serverless site health check system. The “down detector” he made cleverly uses Route53 to watch a site and send a message in Slack whenever an outage occurs. He even provides the full source code in a SAM template so you can get started watching your own endpoints. Big takeaway from me here - Route53 health checks don’t require you to own the site. They will monitor anything with access to the internet! Great article, Darryl!

Tip of the Week

The final weeks of the year are always fun to see recaps from community members. One recap specifically stood out to me - Danielle Heberling. She had a milestone year with some incredible life events and amazing achievements. I aspire to be like Danielle. We all should. Congratulations on an amazing year, Danielle!

New Releases

Amazon Cognito user pools now support the ability to customize access tokens. This has been a long time coming and addresses a huge ask from developers for years!

AWS AppSync added more controls for your GraphQL API usage like max number of resolver executions per query, max depth of data queries, and disabling introspection queries. These are great security/gotcha controls to keep your costs under control.

You can now natively trigger a Lambda function on a CloudWatch alarm state change! This will make it dirt simple to respond to events and keep your app highly available.

Last Words

What a great year! Thank you to all of you, both the new subscribers and the ones who have been with me since the beginning! I look forward to bringing you another great year of catered “best of the best” articles to help us learn quickly and efficiently. Here’s to you all!

If you’d like to make a recommendation for the serverless superhero or for an article you found especially useful, send me a message on Twitter, LinkedIn, or email.

Happy Coding!

Allen