Serverless Superhero

Our serverless superhero this week is Nick Smit, PM for EventBridge at AWS. Nick has had a hand in some of the biggest features we know and love in EventBridge and is an incredibly smart resource when it comes to EDA. Thank you, Nick, for bringing your expertise to EventBridge and literally helping shape EDA as we know it today!

Tutorials

I’m a big fan of the series Jimmy Dahlqvist has been doing with his connected BBQ SaaS app. Last week, he published part 3 of the series, which focuses on tenancy. Jimmy describes a few ways to approach tenancy, then goes into detail about how he built his SaaS app with row-level tenancy. His post covers the APIs necessary to manage tenants (it’s more than 1!) and also goes into auth and CORS. This is a great, practical post that app builders can use directly with what they are working on.

How confident are you when it comes to security in your event-driven architectures? If you’re answer is “Confident…I think,” then you’re like most of the rest of us. James Eastham published a video last week on all things security in EDA. Naturally, he discusses things like thin and thick events, encryption, and callbacks - but what I love about this video is that he relates everything together through an intentional security lens. He explains the concepts masterfully in the video with some on-the-fly drawings and fantastic explanations of everything.

Interesting Content

I did a live coding session with Andres Moreno last Friday where we were trying to get Elastic serverless fully functional in under one hour. The stream starts with Andres and I speculating on the “serverlessness” of Elastic serverless, then we dive straight into coding an API that ingests data and performs a search. We built it on top of AWS Lambda and I was so happy that there were no VPCs used anywhere in the project! The serverless version of Elasticsearch is legit, but still a little pricey. I’ll be curious to see how they change the pricing before it goes GA.

Speaking of search, Ben Pyle did a very similar experiment last week with Typesense. His article shows you how to populate and search a collection in Typesense cloud and how it fits into a serverless architecture. This looks similar to Elastic serverless in functionality, but at about 2% of the cost. While it doesn’t quite fit the “pure serverless” mold many of us like to see, it does land far right on the serverless spectrum, with all the advantages of a managed service. I’ll have to give this one a try myself.

Spotlight

Developers who focus on serverless often talk about ephemeral environments. These are sites that spin up when a PR is opened or a dev is doing some testing and get destroyed when the PR is merged or the dev is done. This works great in apps that only contain serverless/stateless resources. But in many cases we have those nagging stateful resources like OpenSearch or EC2 instances that would take too long to provision or be too expensive to create. Lee Gilmore published a blog post how he approaches the problem. His solution makes sense when you think about it, but you can tell it took a lot of mental energy, blood, sweat, and tears to get there. I won’t spoil his solution here. I highly recommend you see how Lee recommends using serverless ephemeral environments with stateful services.

Tip of the Week

Ben Pyle shared his thoughts on a tweet from David Hansson last week who was making a bold claim stating “if you don’t have over $1M annual recurring revenue for your SaaS app, you don’t need more than 1 server”. Ben stated you could say the same thing for serverless, which is a fantastic point that many people seem to overlook when they opt for containers and managing a Kubernetes cluster from day 1.

I could use this same argument as a case for #serverless. Especially early on. And even more so if you lean into containers as the least common denominator for your portability strategy. But seriously, most don't outgrow serverless and or even Lambda. #BelieveInServerless https://t.co/hd8shLsyFQ

— Benjamen Pyle (@benjamenpyle) August 24, 2024

New Releases

S3 finished rolling out the update to no longer charge for several HTTP error codes. If you remember back in April, there was a bit of a snafu with a DDoS attack resulting in a surprising bill for a user. As of last week, you no longer have to worry about it happening to you.

Lambda released an update to configure resursive loop detection at the function-level. This is a nice quality-of-life update for people who might intentionally be running recursive patterns in Lambda.

Momento is now a partner destination for EventBridge. This simplifies the configuration for async cache updates and message publishing from EventBridge to Momento.

EventBridge Scheduler boosted some default quotas, including number of allowed schedules, invocation throughput, and API request rate. I don’t know about you, but I use the scheduler in most of the work I’m doing these days. It’s an amazing service.

Lambda now supports encryption of filter criteria for event source mappings. This is a nice update for compliance reasons if you’re looking to execute a function given specific data in a queue or stream.

Last Words

What are you working on right now? Are you building something cool and want to share? Or are you stuck on something and want to see how other people tackle a problem? The Believe in Serverless Discord server is a place where you can go to share your thoughts and get answers to questions from some of the best serverless practitioners in the world. Check it out!

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