Good Morning!

I’ll be at the San Franciscos AWS Summit this week. If you see me, say hi! I’m kinda hard to miss…

And of course, last week’s video seems to be a rousing hit; I’m curious to hear what you folks think of it:

<p></p><div class="sc-bxivhb ivbKJD"><a href="" target="_blank" rel="noopener noreferrer"><img src="" style="max-width: 480px;" alt="Click through for source"></a></div><p></p>

From the Community

Are you struggling to determine what analytics workloads can perform well in the data lake, and which ones should be pushed to the data warehouse for peak performance? According to Gartner, you’re not alone. But thankfully, a category of technologies that Gartner calls “analytics query accelerators” are here to help. Get your free copy of the new Gartner Market Guide Analytics Query Accelerators, courtesy of ChaosSearch. Learn how analytics query accelerators provide SQL or SQL-like query support on a broad range of data sources to deliver BI dashboards, interactive query capabilities, and support for data modeling. Help your data lake deliver faster time to value – get the free Gartner report, courtesy of ChaosSearch, today!

When you or I see an article titled When is the Lambda Init Phase Free, and when is it Billed?, we see a fascinating exploration of Lambda. Meanwhile, AWS sees it as a bug report.

I asked Twitter last week whether or not there was a compelling economic case for using AWS (or any cloud provider) for GPUs instead of running them yourself due to the excessively high cost associated with the decision. At least one company seems to agree with me–Enzymit made the same choice and built out a private cloud for this.

I’m not sure what I expected from an article titled Tips for Getting the Most out of AWS CloudFormation on Your Next Project, but "CloudFormation is basically awful, so if you have to use it WHICH YOU SHOULD NOT, here’s how to make it slightly more tolerable" was definitely not it. Hear hear, though!

I asked, in good faith, whether or not there was an economic case for running GPUs in the cloud. The next day TechRepublic wrote about my question. I’m not entirely sure what to make of that…

AWS is apparently building new datacenters in the Bay Area. It would be nice to have a region here! "Well hang on, what about us-west-1?" you may be about to reply, but please; we’re talking about an actual region here, not an expensive capacity-constrained pretender.

Vlad Ionescu has a post on scaling containers on AWS in 2022. It’s worth the read, but he’s missing a few (dozen) ways to run the containers in his analysis.


The Amazon Annapurna Labs team is responsible for building innovation in silicon and software for AWS customers. With development centers in the U.S. and Israel, Annapurna is at the forefront of innovation by combining cloud scale with the world’s most talented engineers. Our team covers multiple disciplines including silicon engineering, hardware design and verification, software, and operations. Because of our teams’ breadth of talent, we’ve been able to improve AWS cloud infrastructure in networking and security with products such as AWS Nitro, Enhanced Network Adapter (ENA), and Elastic Fabric Adapter (EFA), in compute with AWS Graviton and F1 EC2 Instances, in machine learning with AWS Neuron, Inferentia and Trainium ML Accelerators, and in storage with scalable NVMe. Location: While our roles mainly sit in the Bay Area, Austin, and Seattle, a lot of our roles are also open to mid-to-senior hires sitting remotely.

At Modern Treasury, we are building payments infrastructure to power $750 trillion in bank transfers every year. Before Modern Treasury there has never been a universal API into the global banking system. Our ambition is to be the de facto standard for money movement for the world’s most innovative and fastest growing companies. Our customers use our APIs to automate payouts, direct debits, balance tracking and other payments use cases at scale. Join our engineering team at Modern Treasury to help build the new foundation of business and finance.


Last Week in AWS: Denonia Denials

Last Week in AWS: Requiem for a Weasel

Last Week in AWS: Taking AWS Account Logins For Granted

Screaming in the Cloud: Doing DevRel on Easy Mode with Matty Stratton

Screaming in the Cloud: The Anti-Entropy Agent with Johnny Podhradsky

Screaming in the Cloud: Web3 Ain’t All Its Cracked up to Be Molly White

YouTube: Should I Take A Job At AWS?

YouTube: Becoming a Pathfinder in Tech with Emily Kager

Choice Cuts

Software powers the world. LaunchDarkly empowers all teams to deliver and control their software. DevOps and feature management are reimagining how businesses build and release new products. Get control of your code to ship fast, reduce risk, and reclaim your nights and weekends. Learn how your team can reap the rewards of Continuous Delivery without all of the risk. Check out LaunchDarkly.

Amazon EFS integration with the new and improved launch experience on the EC2 Console – "Better late than never" I suppose. The EFS integration in the EC2 console Launch Wizard is a great example of what’s possible when AWS service teams risk their jobs and budgets by violating policy to actually talk to one another.

AWS Fargate now delivers faster scaling of applications – If it were REALLY fast, then wouldn’t it be a Neargate instead?

Introducing the Amazon CloudFront Ready Program – That’s right, the CloudFront Getting Ready Program, which races my toddler as far as "how long will it take you to get ready to go" with respect to latency. Join me on the couch, we’re gonna be here for a while.

Accelerate Snowflake to Amazon Redshift migration using AWS Schema Conversion Tool – Until a post like this is written by an actual customer who’s done it instead of a team of AWS employees, I won’t accept that this is a real thing that customers are doing in any significant numbers. I say this as someone who’s planning to migrate some workloads to Snowflake myself, mind you.

Develop and test AWS Glue version 3.0 jobs locally using a Docker container – It had somehow slipped my mind that AWS Glue 3 was available, but yeah–it came out last August.

Enhanced AWS Backup features for Amazon DynamoDB – "You can now copy your DynamoDB backups across Regions and across AWS accounts, move your DynamoDB backups to a cold storage tier, and add cost-allocation tags to the backups." What you apparently still can’t do is restore a table’s data additively to an existing table without messing around with Glue or similar. Someday, maybe…

Save costs by automating the start and stop of Amazon RDS instances with AWS Lambda and Amazon EventBridge – Save even MORE costs if you’re AWS by not building a scheduling system into RDS itself, instead opting to make your customers all independently do the work themselves, and then charging them the EventBridge plus Lambda fees just to pour a bit of salt into the wound. "Frugality" beats "Customer Obsession" again.

Retailers: Turn Retail Data into Competitive Advantage – Specifically, Amazon’s competitive advantage.

Using Amazon CloudFront with AWS Lambda as origin to accelerate your web applications – I’m incredibly irked that there’s no way I’ve yet found to use a different CDN than CloudFront to reasonably provide custom domains to Lambda Function URLs.

Using a scientific thought process to improve customer experiences – Amazon talking about improving customer experiences via a scientific thought process while having the front of their retail site veritably festooned with sponsored ads and knock-offs of the thing you actually want to buy is exhibiting some "early 2000’s at Google" level of tone-deafness when talking about human interactions. I just cannot take them seriously on the topic, no matter what I do.


While AWS doesn’t like to talk about it, this multi-cloud thing is…well a thing. This is where MinIO comes in. MinIO’s high performance, Kubernetes-native object store works on every cloud – literally all of them from AWS to Zayo. This means you can build S3-like data infrastructure anywhere. The world’s fastest object store with READ/WRITE speeds in excess of 325 GiB/sec/165 GiB/sec respectively, MinIO can handle any workload – from modern databases to AI/ML and advanced analytics. Couple that with a suite of enterprise features for ILM, IAM, security and resilience and organization can architect consistency for their data persistency – across and between clouds. Don’t take our word for it, see for yourself at

ts2asl is a TypeScript to AWS ASL transpiler. I’d suggest that Stedi is soon to pivot away from EDI solutions and into building serverless developer tools except that I kinda suspect it happened a while ago. They do a LOT of great work like this over there…

I’m not quite sure how to take it, but my article on Taking AWS Account Logins For Granted has led to an explosion in the Granted ecosystem, with people showing up left and right to contribute patches. YES! Open Source is working, and I didn’t even have to implement Amazon Basics Granted to achieve positive change!

… and that’s what happened Last Week in AWS.

Newsletter Footer

Sign up for Last Week in AWS

Stay up to date on the latest AWS news, opinions, and tools, all lovingly sprinkled with a bit of snark.

"*" indicates required fields

This field is for validation purposes and should be left unchanged.
Sponsor Icon Footer

Sponsor a Newsletter Issue

Reach over 30,000 discerning engineers, managers, and enthusiasts who actually care about the state of Amazon’s cloud ecosystems.