Serverless Superhero
Our serverless superhero this week is Tycko Franklin, senior software engineer at Pacific Northwest National Laboratory. Tycko is a very active member of the serverless community, offering his insights on community issues and sharing content from his fellow creators. Thank you for everything you do, Tycko!
Tutorials
The Rust hype train is picking up momentum every week. James Eastham shared a video with us last week on the perfect structure for production-ready serverless Rust projects to help answer the important non-functional requirements of software. He covers setup using the CDK and talks about API projects and single-purpose Lambda functions as well.
Wait, did someone say Rust? Ben Pyle continued with the Rust content with his post on DynamoDB streams and Rust. His article shows you how to set up the infrastructure using TypeScript CDK (I know, right?) and how to process DDB records in Rust. Naturally, he explains everything that’s happening in the handler and shows incredibly low latency of ~2ms when processing 100 records off a stream! I have to admit, I’m impressed.
Lee Gilmore wrote an article last week on Lambdaliths and function urls that quite frankly both caught me off-guard and made me very excited. He describes the benefits of putting an entire API inside of a single Lambda function and accessing it through a function url. He, of course, covers auth and gives a practical example of how you can get started with this pattern. I really like this idea and it fundamentally changes the way I’ve thought about and approached API design in serverless environments.
Interesting Reads
Have you ever seen the init duration time and duration time in your Lambda logs and wondered what the difference is? If you add them up, it’s not the total execution time - so what does it represent? David Behroozi explains the difference and details out for us the timeline of a Lambda function execution in his post on the Lambda request timeline. This is a really nice read in a FAQ format with meaningful pictures. Great explanation that can help you get a better understanding of what’s actually happening when your code runs.
It’s almost like Lee Gilmore and Ryan Cormack coordinated efforts last week. Ryan wrote an article comparing the performance of Lambdaliths and thin functions. This is less on the architecture and more focused on the cold starts and invocation time. I highly recommend reading both articles when weighing your decision of how to approach your architecture.
Podcasts
I had Jeremiah Dunham on the Ready, Set, Cloud podcast to talk about math in the world of computer science. Specifically, we spoke about automated reasoning and how there’s still a need for advanced mathematics in a world taken over by Generative AI. It’s a fascinating conversation and Jeremiah hints at a possible fundamental change in how we test our code in the not-too-distant future.
On the Real-World Serverless podcast, Yan Cui had Mark White on to discuss a case study of moving from on-prem to serverless. The episode focuses on Mark’s place of work, Dunelm, and the challenges they faced along the way from engineering culture change to technology limitations.
You might have heard of LLRT, low latency runtime, that AWS recently released for Node.js. On the AWS Bites podcast, Eoin Shanaghy and Luciano Mammino talk it over and give us everything we need to know about it, from implementation details to benchmark comparisons.
I was on the PodRocket podcast last week talking about why you should stop building synchronous applications. This is a replay of a talk I recently gave at THAT conference where I dive into approaches to keep users hooked while async processes are running. I talk about interactivity, why that’s important, and how serverless really lends itself well to this paradigm.
Spotlight
Real-world case studies fascinate me so much. Nothing speaks truth like production battle wounds. The #believeinserverless community had Marcin Sodkiewicz on the live stream last week talking about RyanAir’s journey into serverless. He covers the designs they’ve worked through to handle massive scale and even covers some innovative solutions they’ve built to cut costs tremendously.
Tip of the Week
If you’ve been putting off building a GenAI app, it might be time to try it out. Marcia Villalba consolidated all the resources she’s created on the topic in a single blog post, giving you a full set of tutorials and explanations on exactly what to do.
New Releases
Amazon Bedrock Knowledge Bases now support Aurora PostreSQL and Cohere embedding models. AWS seems to be going all in on Knowledge Bases for Bedrock, which will help make tailored experiences for your end users.
AppSync has been on a tear lately. They just released 12 new CloudWatch metrics for enhanced monitoring including request/error counts, latency, and cache hits/misses.
Last Words
Wow what a busy week! Lots of podcasts and videos are making their way to the top of the community. I’m sensing a shift in the way we’re consuming content these days and I’m wondering if you’re changing your habits of how you learn. Are you leaning more toward other mediums this year? Let me know!
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