I'll deploy it through the Serverless Application Repository, meaning I'm now a full third of that thing's customer base for October.
Now I need a Lambda function to test this on. Fortunately I have one that fires off once a minute, is idempotent, and powers the horrible yet strangely compelling @awscwoud.
Let's give this config a shot. (Somewhere an AWS employee is looking up the account ID and discovering that the globally unique account alias for this @awscloud account is indeed "Shitposting.")
And we're off. Now I wait a smidgen.
There we go. No surprise, "cheap" or "good" are the options here. Now let's see if this asplodes when I kick the Lambda over to use Graviton2.
I use the advanced DevOps IaC approach of "using the @awscloud console, then lying about it" to make the change. Yes, I really did call my handler that.
(Had to repackage and re-upload it just because I'm a sensible person who sets code packages to expire.)
Now we run the thing and see if it melts. Hope with me!
Excellent job hoping. Let's see.
I don't know about you, but looking at five zeros before a number doesn't help me get a sense of cost overly well. Fortunately @alex_casalboni has thought of this with the handy "Compare" button.
It looks like 256MB + Gwaviton2 is the most cost effective answer here unless I misread the admittedly complex chart. Am I right?
John is completely correct, but remember that this is a backend job. The latency absolutely does not matter; @awscwoud doesn't need to respond THAT quickly.
An update on my Graviton2 Lambda conversion. The most obnoxious part was getting the tooling updated to support the new configuration. The wrapt library by default compiles a C extension; you can override it to use pure python, but that's not straightforward.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Hello, and welcome to our company's oh-so-very-shitty Security Awareness Training. I'm Chief Cloud Economist Corey Quinn of the Duckbill Group, and I'll be delivering this training for you because I was absolutely NOT the lowest bidder for a change.
The whole point of security awareness is to protect company information. That's what they say, anyway. Here in reality we're going to reference back to the things I spew at you rapid fire and blame you for our institutional shortcomings once we get breached.
Confidentiality is important. Assume that people will read what you write. I know, it's a heavy lift for some of you who haven't figured out that the failure mode of "clever on Twitter" is "being a huge asshole," but pretend it'll be read.
Let me begin by saying I'm *pretty* sure that this isn't sour grapes; I left ten years ago as of six weeks from now. I have no financial stake in Expensify, but I do have stories. I was the "Director of TechOps" for a year.
Let me further disclaim that my information is thus a decade old. I hope it's all ancient history, but these stories have a way of helping people out. I hope and trust that as we all have, the culture there has evolved significantly.
First, the good parts. I've never seen a place that was as good as Expensify was about onboarding junior engineers and training them basically from scratch. "Senior engineers are expensive, so we'll hire them before they get there" is a great philosophy.
You can fork the SSPL or Elastic license or what have you to only include @awscloud, but it's 1. not Open Source® so people will yell at you for it, and 2. it betrays a significant misunderstanding about how AWS works.
Let's pretend that we're talking about `cowsay`, an open source tool that's been around for decades, is written in Perl, and outputs text in terminals like this.