Serverless Superhero

Our serverless superhero this week is Filip Pyrek, founder of Buttonize and AWS Serverless Hero. Filip co-organizes the serverless Brno Meetup group and has been highly active in the serverless community for years. He currently has focus on simplifying internal company operations with his beautiful app, Buttonize. He’s provided a secure, intuitive way to directly call Lambda functions and other AWS services from an easy-to-navigate user interface. Thank you for everything you do for the community, Filip!

Tutorials

Benjamen Pyle has been on a roll lately. In his latest blog post, he describes a pattern he calls the “Event Bus Mesh”. This pattern connects Event Buses together to promote data and compute isolation between microservices. I’m all for data boundaries and the decoupled nature of event-driven architectures. Patterns like this prevent you from accidentally connecting services with strong references to resources, which could end up resulting in the dreaded “distributed monolith”. Benjamen’s article comes with functional CDK so you can get started with the pattern on your own.

Developer experience should be top of mind for anyone building an API or SDK intended to be used by a 3rd party. I’ve been building a game recently and had an epiphany that helps simplify APIs drastically. I (Allen Helton) wrote a post on how to implement user sessions to simplify APIs. By storing session-based data in your API, you can simplify paths, boost performance, and improve DX. Not all APIs fit in a user session-based model, but for the ones that do - you have a great opportunity to enhance your DX.

Interesting Reads

One of the most interesting articles I’ve read in a long time is from AJ Stuyvenberg. He uses a new capability from Datadog to dissect the bits and pieces of a Lambda function cold start to figure out where the latency comes from. He runs multiple experiments to compare cold start times, like using the AWS SDK v2 compared to AWS SDK v3 for JavaScript. He also shows the impact of bundling resources with webpack or esbuild compared to unbundled. Most importantly, he explains the results in a meaningful way that takes the mystery out of cold start times. Extremely cool analysis!

While we’re talking about observability, Thomas Ankcorn walks us through an invaluable tool we don’t use enough - CloudWatch Insights. Thomas goes through the capabilities of the log parsing service, showing us some of the true power it holds when gathering insights or debugging issues.

We all know software is built via a sequence of trade-offs. Davide de Paolis explains in his article why having several options to solve a problem is so important when determining trade-offs. It’s not always about getting it done the fastest, but often it’s about getting it done the most correct. My takeaway is that there is no such thing as perfect software, and spinning your wheels trying to make the perfect decision will lead you straight down a path to failure.

It can be difficult to tell a serverless application to slow down. They scale so quickly when bursts of traffic come in, sometimes downstream services choke trying to keep up. Yan Cui wrote a post on 3 ways to manage concurrency in serverless applications to address this problem. He talks about reserved concurrency, event source mappings, and how you can take matters into your own hands with dynamic concurrency control. These are great recommendations that get you thinking a little differently.

The videos were released from ServerlessDays Belfast last week. There’s some amazing content on there from phenomenal speakers, with most of the videos around 15 minutes in duration. Worth catching up on if you weren’t able to make it out!

New Releases

SNS now supports setting content-type request headers for HTTP notifications. This eliminates assumptions in subscriber handlers for content type, making for an easier and more predictable workflow.

Spotlight

Michael Walmsley wrote a wonderful article on serverless scalability boundaries that perfectly summarizes why this is the year of observability. He talks about the importance of flow control, reminds readers that production workloads generate higher amounts of traffic and often different results than controlled testing environments, and describes how to resolve issues with scalability clashes. Funnily enough, this is one of those problems that arise because serverless services sometimes scale too well.

Podcasts

Episode 6 of the Ready, Set, Cloud podcast released on Friday. I talk with AJ Stuyvenberg about why serverless observability is so hard and what the future holds to make it simpler.

Real World Serverless released an episode with Kirk Kirkconnell talking about the litmus test for serverless. They talk about caching, single table design, and how to spot a faux-serverless service from a mile away.

A cool episode of AWS Bites with Luciano Mammino and Eoin Shanaghy came out last week answering the question, is AWS App Composer any good? They do a great job describing the service and weighing in with their opinions.

Tip of the Week

Luc van Donkersgoed shared a quick glimpse of the 2023 AWS survey results. It’s a cool “state of AWS usage” summary to see how you stack up with others in the industry.

The results for the 2023 AWS survey are in! Lots of interesting insights, like:
- More people use Serverless Framework than CDK
- More people use SNS than SQS (surprised me!)
- GitHub Actions has 99% retention (!)

All results 👇https://t.co/XsWZ8h6UBV pic.twitter.com/KLcdtBEmMG

— Luc van Donkersgoed (@donkersgood) March 16, 2023

Last Words

This is Serverless Picks of the Week’s birthday! For those of you who have been along for the ride, thank you. For those of you just joining, welcome and I’m glad you’re here!

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