Before we get into the AWS-specific stuff, I want to highlight two friends who’ve done what I’ve never dared: written books. Ana Visneski’s F*ck It, Watch This- Saying the Quiet Parts Outloud is about… well, basically my entire career strategy if I’m being honest. Meanwhile Gergely Orsosv’s Software Engineer’s Guidebook is almost certainly a better approach than my own "finding the sharp edges via flailing around in the dark." They’re both books you want to pick up.
It’s officially pre:Invent season, or as I say: pre:Quinnvent season, and we’re gearing up for it. Don’t miss our Wednesday night drinkup at Atomic Liquors or the Nature Walk! I think you’ll be pleasantly surprised by how much fun those two events can be…
From the Community
This Supercharging a Serverless Slackbot with Amazon Bedrock is really neat; unlike the official Bedrock examples they don’t drag in WAF or Kendra–so this is super inexpensive rather than costing a majority in "non-Bedrock services." It’s affordable, and one of the first things I found that helped me "get" what Bedrock is aiming to become.
The Hidden Visibility of AWS Service Limits has been driving me up a wall for many, many years. The three kinds of AWS service limits / quotas are "soft," "hard," and "the ones AWS Support swears don’t exist for three rounds of ticket tennis." Trusted Advisor’s dismal coverage percentage just shore up my position that it’s a failed service that should be Googled.
Fumbling with AWS is a very relatable tale of someone who’s familiar with computers starting out with AWS in 2023. This is a core problem inherent to AWS services: they don’t get to see the beginner perspective very often. And man, does the "stitch together 40 blog examples into a Frankenstein’s monster of a codebase" resonate with me.
Philipp Schmid, Technical Lead at HuggingFace and AWS ML Hero, penned Amazon Bedrock: How good (bad) is Titan Embeddings?, and I’m paying unusually rapt attention to it. Honestly, with AWS’s "we are the best at Generative AI" hogwash, nothing they say about the space can be believed at this point–and that’s both novel and profoundly uncomfortable for me. It therefore falls to experts like Phil to tell us the straight dope about AWS’s fledgling offerings…
Last Week In AWS: C-Suite Responsibility
Last Week In AWS: High Cardinality Service Usage
Last Week In AWS: How to Stop Feeding AWS’s AI with Your Data
Screaming in the Cloud: Building a Strong Company Culture at Honeycomb with Mike Goldsmith
Screaming in the Cloud: Learnings From A Lifelong Career in Open-Source with Amir Szekely
Amazon Aurora Global Database for PostgreSQL now supports write forwarding – If your architecture doesn’t require strict transactional ordering, this potentially opens the door for what’s basically going to feel like magic. I mean, I wanted this from regular RDS back in 2014, but better late than never.
Amazon SQS announces support for JSON protocol – In a radical turn of events, AWS has decided to propel itself into the 21st century by adding JSON protocol support to Amazon SQS. Hats off to Amazon for boldly embracing a technology first specified in 2001. I’m sure that soon they’ll wow us all by discovering fire or inventing the wheel just before re:Invent 2028.
AWS Cost Management now provides purchase recommendations for Amazon MemoryDB Reserved Nodes – I saw this come out and, as is common, thought it had been there for a while and I’d just missed it historically. I still don’t see widespread use of MemoryDB, but at least it’s a bit easier to use it economically if you’re in an unfortunate spot?
Introducing the Generative AI Center of Excellence for AWS Partners: The Path to AI Expertise – If this isn’t simply a redirect to talk to either OpenAI or GitHub, it’s the wrong answer. Asking AWS for advice on AI is like asking for tips on long distance running from someone who faceplanted three steps in to a 10K race.
New – Block Public Sharing of Amazon EBS Snapshots – I am once again asking who in their right mind would ever think that sharing a raw copy of a disk volume would be something that they’d want to do casually, let alone at all.
New for Amazon Comprehend – Toxicity Detection – There are a non-zero number of human AWS employees who seem unable to distinguish "me making fun of a service" from "personal attacks on individuals," so you’ll forgive me if I’m extremely skeptical that they’ve managed to build a robot that can thread that needle of nuance.
AWS CodeBuild adds support for AWS Lambda compute mode – CodeBuild remains the best serverless offering that AWS has to run containers in certain circumstances, and this makes that offering even better. The best global offering in this space of course remains Google Cloud Run, assuming it doesn’t get cancelled between the time I write this and the newsletter getting sent.
An Overview of Bulk Sender Changes at Yahoo/Gmail – If only AWS could give us useful overviews like this of changes and enhancements to their own services. I guess I need to ask Microsoft for that kind of roundup as this ridiculous industry builds its own duckloop of information dispersal.
Creating a correction of errors document – This is a great guide on how to do something that (based upon what they put in public) AWS itself does one time out of every dozen or so outgages.
Know Before You Go – AWS re:Invent 2023 | AWS Management Console – "Download the AWS Console Mobile Application and configure notifications so you don’t miss important events" says the post, competing at an elite level of Skillfully Dodging the Actual Point of What They’re Talking About. Suddenly I am extremely worried about exactly what the hell they’re going to be announcing and how. "It’s time for a big announcement so we’re going to take down us-east-1 like it’s the Apple Store website to make sure everyone’s paying attention" is going to kinda put a crimp in my week.
Unhoused individuals gain shelter, prove their identity using AWS-powered solution Kiip – It’s very on brand for AWS to take a look at the unhoused tragedy, see a wide breadth of areas that need help–and then go leaping to solve the "papers, please" part of the story.
How VMware partnered with AWS to nurture a culture of sustainability – Oh, so AWS helped build VMware’s current culture? Please, go talk to any current VMware employee about how the work culture is these days and tell me how that goes.
My post on How to Stop Feeding AWS’s AI With Your Data has gotten a lot of "hey, we didn’t know that was happening; we’re now in panic mode" responses from a number of customers. Conversely, this may be the first post about AWS I’ve ever written where someone from Amazon didn’t reach out to me to discuss it. One wonders.
termshot is a fun program I was introduced to at GitHub Universe last week; I’m playing with it for terminal screenshots.
… and that’s what happened Last Week in AWS.