Off-by-none: Issue #247

September 19, 2023

Only have a few minutes? Check out this week's MOST POPULAR links as chosen by our email subscribers.

Stop taunting me, OpenSearch Serverless! 🤨

In our previous issue, we looked at Bun’s potential impact on serverless, met some new heroes, and learned to stop thinking about old compute. This week, OpenSearch Serverless gets more serverless, SNS FIFO topics get less picky, and AWS tries to stop hallucinating. Plus, we have some awesome serverless content from the community!

Serverless News & Announcements 📣

There wasn’t any major AWS serverless news this week, but plenty of incremental improvements worth mentioning. First up is the API Gateway console refresh. I mean, I guess it’s better. But at least the dark mode is nice. 😐

AWS AppSync WebSockets-based subscriptions for real-time updates support nested filtering now, which is a nice little feature to limit what gets published to subscribers, Amplify Studio now offers full support of GraphQL APIs, giving developers more data options, and Amazon SNS FIFO topics now support message delivery to Amazon SQS Standard queues. This last one is actually quite useful, opening up some new patterns that can mix and match ordering requirements. Also, according to Heitor Lessa, Powertools for AWS Lambda (Python) 2.25 is out.

Amazon OpenSearch Serverless extends auto-scaling to handle high query rates, which allows it to now “efficiently handle tens of thousands of query transactions per minute.” So you can “rely on OpenSearch Serverless to help handle unpredictable surges in your search and query traffic.” PLEASE MAKE THIS SERVICE MULTI-TENANT AND OFFER IT TO ME LIKE DYNAMODB TABLES! Sorry for shouting, but this would be a game-changer.

In “AWS is now a GenAI company” news, apparently Pinecone is working with AWS to solve Generative AI hallucination challenges. Sounds interesting, though I’m pretty sure this press release was hallucinated by GenAI in the first place. Plus, you can now Connect Foundation Models to Your Company Data Sources with Agents for Amazon Bedrock. Context, context, context is so important with these models, so this is actually quite cool (probably).

Last week we talked a lot about Bun and serverless, and the buzz is still going. Yan Cui asks and investigates, “Is Bun the Next Big Thing for AWS Lambda?” Also, Maciej Radzikowski had an X/Twitter thread with some hot takes as well.

Serverless Concepts 🏗

Enhanced Error Handling for Step Functions Provides Developers with Fine-Grained Control over Retry
Steef-Jan Wiggers does a recap on some responses to the new error handling capabilities of Step Functions. #jitterforthewin

Monitoring SQS with Datadog
Benjamen Pyle shares some insights (and code) showing you how to monitor your serverless queues by instrumenting with the proper observability tools.

Load Testing Serverless Application using k6
Marko says “serverless is ultimately scalable until it is not.” Which is very true. See how to make sure your patterns do scale by using k6.

Change Data Capture with Amazon DynamoDB using DynamoDB Streams and Kinesis Data Streams
Two CDC patterns and the use cases they support. The number of consumers is an important factor, though I’d say unless order is strictly required for lots of consumers, events might be a simpler solution.

Stop Abusing DynamoDB Transactions. Leverage Idempotency Instead
Good solution from Matteo Agius here. I’ve been using this type of pattern for quite some time, and Matteo does a nice job explaining it.

Serverless Tutorials 👷‍♀️

There were lots of great serverless tutorials worth sharing this week. Here is a sample of some of the more interesting ones.

• Serverless Setup in AWS with SQS, SNS, API Gateway, & Lambda by Chinelo Osuji (set aside a few hours for this one 😉)
• Bun + AWS Lambda = ❤️ by Vargha Laszlo
• How to send transactional emails with Resend and Ampt by Emrah Samdan
• Getting started with search on a Vector Index by Ellery Addington-White
• Building Typed Serverless APIs With tRPC
• How to Securely Pass Secrets to Lambda | AWS CDK 😎 by Florian Sabani
• Building a StackOverflow Clone | Part 2 by Binh Ngo
• How to Pass JWT Tokens Sent in an HTTP Header Request in AWS API Gateway to Step Functions by Alex Gmoca
• Deploy AWS AppConfig Configuration with the New L2 CDK Constructs by Shashank Kumawat
• Implementing access control on API Gateway endpoints with ID tokens by Arpad Toth

Serverless Reads 🤓

I love serverless, but it got something dramatically wrong
I had a rant on X/Twitter as to what I thought was one of the major issues with serverless. Upon request, I turned this into a blog post. I’d love to know your thoughts.

3 critical stops on the back-end developer roadmap
Probably nothing particularly new to learn here from Darcy DeClute’s piece, but the theme here is that you need to learn Cloud Patterns (which including message-based systems, i.e. EDAs, and serverless). Learning the patterns, IMO, is more important than learning the implementation details.

Scaling With Serenity: How DHARMA Streamlined Operations with Ampt
Emrah Samdan shares how Ampt reduced operational complexity and empowered DHARMA’s small engineering team to get to market quickly.

There are lots of ways to put a database in the cloud – here’s what to consider
Lindsay Clark’s post is essentially about specialization (around databases), and puts some context around what we’re expecting from serverless here.

How to Choose the Right Architecture for the Right Job
It still amazes me how many people fundamentally misunderstand the Prime Video thing, including the author. Luckily the Q&A with the speaker the article promotes holds much smarter positions.

Podcasts, Videos, and more 🎧

Choreography vs Orchestration for serverless applications
One of my favorite topics in distributed system design. I’ve written extensively about this as well, but Yan Cui has a short video that lays out the most important bits.

Kinesis Data Firehose with AWS CDK – Lambda transformations
Marcia Villalba shows you two different configurations to create a Kinesis Data Firehose with AWS CDK that puts records into S3.

AWS Bites Episode #95 – Mounting S3 as a Filesystem
Eoin and Luciano explore the pros and cons of AWS S3 Mountpoint and deliver some solid reasoning as to why it might not be the right abstraction.

Building an API platform on AWS | Serverless Office Hours
Giedrius Praspaliauskas joins Julian Wood to go through a new AWS guide to cover the end-to-end needs of the API lifecycle.

New from AWS 🆕

Here are some other bits of AWS news you might have missed:

• AWS Glue Streaming now supports Kinesis Data Streams enhanced fan-out feature
• Knowledge base for Amazon Bedrock connects foundation models to your data sources
• Cost Anomaly Detection increases custom anomaly monitor limit to 500
• Amazon Redshift announces cross-region data sharing support through AWS Lake Formation
• AWS Identity and Access Management provides action last accessed information for more than 140 services
• Amazon CloudWatch Synthetics announces new Synthetics Python runtime version 2.0
• AWS Marketplace introduces new self-service listing feature for SaaS products

Thoughts from X/Twitter

The CDK was a step in the wrong direction. Just sayin’. ~ Me

I had a spicy take the other day on why I think the CDK was the wrong approach. There was a lot of (mostly) thoughtful discussion in the comments that are worth reading. Lots to learn here.

Upcoming Serverless Events 🗓

September 20, 2023 – Ampt Launch Day!

September 21, 2023 – ServerlessDays Cardiff

September 23-24, 2023 –ServerlessDays Tokyo 2023

September 29, 2023 – CDK Day

October 20, 2023 – ServerlessDays Cape Town

October 26, 2023 – EDA Day Nashville

October 26, 2023 – Serverless Summit Hamburg

November 27 – December 1, 2023 – AWS re:Invent 2023

Serverless Star of the Week ⭐️

This week’s start is Maxime David (@_maxday). Maxime is a Software Engineer at Datadog, a self-proclaimed serverless addict, and an AWS Community Builder. One of the coolest things he’s done is create lambda-perf, which provides Lambda Cold Start benchmarks for 28 different runtimes, including recently added ones for Bun. Maxime is also a speaker and YouTube creator. Thank you Maxime for being such an amazing contributor to the serverless ecosystem!

Final Thoughts 🤔

This is a big week! With the upcoming launch of Ampt tomorrow, our team has been insanely busy getting prepped. Please make sure to check out the livestream tomorrow and help us revolutionize serverless app development.

See you next week,
Jeremy

I hope you enjoyed this newsletter. We’re always looking for ideas and feedback to make it better and more inclusive, so please feel free to reach out to me via X (formerly Twitter), LinkedIn, or email.