Good Morning!
I’ve mentioned things a few times in the intro that people who vociferously devour the newsletter claim to have missed entirely. It’s true; people never read as much as you think they do, which is why this feels like a very safe space to admit that 91 years ago it was I who kidnapped the Lindbergh baby. Also, AWS released some stuff last week; let’s dive in!
Also:
I’m performing a live Q&A next month; submit your questions here!
From the Community
Tailscale Funnel makes it easy to share your local development over the internet for collaboration, testing, and experimentation. With Tailscale Funnel you can receive a webhook from GitHub, share a local service with your team, or host a status page from your own computer.
Funnel securely exposes your dev environment at a stable URL, complete with auto-provisioned TLS certificates. Use it from the command line or the new VS Code extension.
I’m on the fence about the idea that you should squeeze the hell out of the system you have. I think it’s highly contextual, and computers are less expensive than engineering time. That said, there’s clearly a point in cloud where scale outweighs the engineering efforts involved. Like anything, it’s tradeoff-driven.
Amazon is "racing to catch up in Generative AI with custom AWS chips", but it’s also "three steps in to a 10K race" and "we’re not behind," and "every Amazon team is working with Generative AI." Please ask the Generative AI that runs AWS Marketing’s messaging efforts to stop hallucinating.
My analysis of The Amazon Prime Day 2023 AWS Bill has been highly amusing to a number of AWS folks. Thanks; I’m here to help!
This is a thoughtful analysis on the death of the AWS Go Lambda Runtime.
Choice Cuts
AWS HealthOmics supports cross-account sharing of omics analytics stores – I’ve often noted that AWS makes up words–in this case, "Omics." I’ve also observed that Amazon’s approach to industry verticals is to take the same thing they sell everyone else and slap the industry vertical’s name onto a version of it–in this case, "Health." And so, here we are, with something called "AWS HealthOmics" and someone dearly needs to stuff the Generative AI responsible for apparently everything at Amazon these days back into its cage, because this is ridiculous.
New – Amazon EC2 M7a General Purpose Instances Powered by 4th Gen AMD EPYC Processors – The price/performance improves from the m6a, but per instance these cost a fair bit more. Be warned.
Amazon OpenSearch Serverless expands support for larger workloads and collections – Cool, now how about it expand support for being actually serverless? You’ve gotta manually provision capacity on this thing, which is basically just OpenSearch with extra steps.
Reduce Lambda cold start times: migrate to AWS SDK for JavaScript v3 – Or eliminate them entirely by migrating to EC2, so long as "Migrating" is a mitigation that’s on the table.
Architecting for Resilience in the cloud for critical railway systems – Uh… am I the only one who’s a bit concerned at the idea of life-critical systems hanging out in us-east-1?
How Amazon Shopping uses Amazon Rekognition Content Moderation to review harmful images in product reviews – Witness Amazon’s leadership and prowess with AI/ML given one of its showcase uses: enforcing Amazon Product Reviews’ sterling reputation for honesty, accuracy, and unimpeachable integrity.
Zero-shot text classification with Amazon SageMaker JumpStart – I need to drink at least six shots before I’m willing to try out anything labeled "SageMaker" again, seeing as how they still owe me $260.
Build a multi-account access notification system with Amazon EventBridge – Way too much work for something that should frankly be a default for new AWS Organizations, if you ask me. Someday the AWS product owners are going to have a revelation: perhaps customers have goals they’re trying to achieve that aren’t "the care and feeding of their AWS estates," and that they might get further faster if they helped customers achieve that. This solution? It’s the textbook definition of "undifferentiated heavy lifting," made all the worse because it’s lifting imposed by AWS itself.
Getting Started with CloudWatch agent and collectd – I’m sorry, why can’t CloudWatch Agent do this itself again? If I have to install a monitoring agent on my nodes, I’m doing it once; I’m not installing a whole suite of observability tools.
Addressing gender inequity in the technology industry – I went digging for Amazon’s own gender equity numbers to set up a "this you?" moment, but was pleasantly surprised to discover that they’re trending in the right direction across the board. Yes, there remains work to be done–but they’re clearly doing it.
Tools
I found a throwback to a comment I’d made years ago; AWS Tags As A Database (AWS TaaDb) is completely free and does the horror you’d expect.
I forget that some folks still get to play with AWS through a lens of Windows; here’s a guide on how to add the AWS CLI to Windows Terminal.
… and that’s what happened Last Week in AWS.
