Our serverless superhero this week is Matteo Depascale, cloud architect at Storm Reply and AWS Community Builder. Matteo is a great blogger, frequently posting articles on serverless, generative AI, and blogging. Thank you for all you contribute to the community, Matteo!
If there’s one thing I love most about developers, it’s the uncanny knack they have for finding workarounds to problems. Zied Ben Tahar found an interesting way to use Step Functions distributed map with S3 buckets in other AWS accounts. This functionality isn’t natively supported, so he had to come up with a way to do it on his own. He explains the solution, details the potential issues with his approach, and offers the source code.
I’m a big fan of Pubudu Jayawardana’s writing style, so I was super happy to see he continued his series on retrying failures with SQS last week. In part two, he describes a way to use EventBridge Scheduler to redrive failures. This is a super cool idea if you need to backoff for an time of longer than 15 minutes. Good read!
Sam Williams released a video with Sam Lock about how he approaches serverless architecture design. The two discuss some of the big serverless topics, like cost at scale and vendor lock-in, in addition to getting into the weeds of design.
As we all know, we can’t go a couple hours without talking about AI these days. And I kinda like that. Especially when we get gold from people like Lee Gilmore who show us how to use it to generate Mermaid diagrams for all types of use cases. Lee talks about several practical types of Mermaid diagrams and shows the prompt, output, and rendered diagrams that you can get in seconds. So cool!
If you were looking for a way to use the new DynamoDB incremental export feature with Step Functions, you’re in luck. Benjamen Pyle wrote a blog post for you. He details how he tracks points in time for the run, sets up the async job, and how he waits for processing to complete. This is a great solution for roll-up or batch processing if you don’t need to track the changes in real time.
SQS FIFO queues now support up to 18,000 TPS for high throughput mode. The team is making a big effort to get throughput parity with standard queues. Note - this new limit does not apply to all regions.
AWS announced the European Sovereign Cloud last week, which can be thought of as the EU version of the United States’ GovCloud. Super happy to see this coming.
Ampt released about the simplest abstraction on top of AI I think I’ve ever seen. They provide you easy access to Amazon Bedrock with minimal code and intuitive commands. If you’ve been waiting to try Ampt, now is the time.
The Serverless Framework is releasing V4 and will begin charging customers for usage, among other things. Their target is companies with over $2 million in annual recurring revenue (ARR) and aim to not charge for things like ephemeral environments. I totally understand the need to make money, but the announcement felt like they were too afraid to commit to a pricing model. I’ll be curious to see how it shakes out and how prices change as they become more comfortable with charging customers.
The internet was buzzing last week with a post from Elias Brange on how he used Momento Topics to test event-driven workflows. This is a really clever solution where he captures EventBridge traffic and pushes it down to his test runner to “synchronously test asynchronous workflows.” It’s a great idea that really opens a lot of opportunities for testing EDA.
EDA day was last week and I had the privilege of being one of the speakers. It’s a humbling experience being around some of the best in the world in modern computing. The videos of the talks will be up soon, I’ll post links when they are live.
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
Thank you for subscribing!
View past issues.