Our serverless superhero this week is Heitor Lessa, chief architect of powertools at AWS. If you thought Lambda was great, wait until you see the magic behind Heitor’s masterpiece. Powertools is an open source project, so Heitor won’t take all the credit - but credit where credit is due. Thank you for everything you do for the serverless community, Heitor!
You may have seen that Step Functions recently released better error handling in workflows. But all great features are useless until we try them out. Jones Zachariah Noel published an article on how we can use the new jitter functionality to retry errors in our workflows. Turns out, the Step Functions team made it super simple to implement. Jones shows the one-liner you need to add and gives some benchmark details on what he saw.
I did an experiment last week that resulted in a super cool, practical tool set. I made a GitHub action that converts your REST APIs to EventBridge events. It takes an Open API specification and generates a CloudFormation template with EventBridge API destinations and rules to trigger them. It’s a great, low-lift way to add a “fire and forget” set of triggers for your API.
There’s no right or wrong runtime to build your Lambda functions in. That said, it’s always interesting to hear about migration stories from one to another and understand the reasoning behind then. Marcin Sodkiewicz told us about his team’s journey going from TypeScript to Go. He talks about the migration strategy a little bit but the big wow factor is his direct comparisons between the two runtimes. He shows how the latencies and costs changed when he finished making the switch over to Go.
An interesting case study was published by Sam Hardern last week. He wrote about how his team restructured their serverless app structure because of scalability concerns. It’s an interesting read as he describes the decisions and outcomes. I went through something like this personally a couple of years ago and while I don’t necessarily advocate for everything he did, lessons will be learned that will make for a stronger future.
I wrote another article last week highlighting some of the cool builds you can do in the front-end with Momento. I (obviously) love serverless services and it feels like an entirely new world when bringing them to the UI. I haven’t been a web dev for too long but this already feels like a game changer to me.
I had Ro Ndimofor on the Ready, Set, Cloud podcast last week to talk about cloud certifications. Ro takes a hard-hitting stance against them and offers some recommendations for how you should spend your time instead. This is a real thought-provoking episode!
A super nice feature was released by Baselime last week that emails you when error logs pop up in CloudWatch. I have already discovered several bugs in background processes that were silently failing because of this. If you’re already Baselime user, you have nothing to do - the feature is already enabled.
I stumbled upon some gold last week. Marko published an article on how he uses types from his Open API spec all the way down the stack. With a few key tools, Marko is able to define his types once and build an API definition from them, validate input, and create instances on the fly. Extremely cool project with tons of practicality!
I make it no secret that Step Functions is my favorite AWS service. I was super excited to see this video from Yan Cui last week where he explains why testing workflows are hard and what he does to get around some of the awkwardness. This is a great video with some priceless information!
🤔 Why is testing Step Functions so darn tricky?— Yan Cui (@theburningmonk) September 21, 2023
Let me break it down for you in my latest video. And I will let you in on my strategy for testing my state machines 👇
There are conferences going on all around the world right now. Are you attending any? Did you see a particularly helpful talk? While I try to stay up to date, I miss a few things every now and then. Let me know of anything you saw that you’d like me to share!