Building observability to increase resiliency

Without question this talk was my most ambitious art project so far. I attempted perspective and depth, added animations to show the ship rocking, set up multiple scenes from the below-deck cargo store, to overboard drills, to radar terminals and zoomed out maps. All graphs in the talk were hand-drawn too, which was surprisingly tedious.

I had a hard time figuring out how to get a continuously looping animation of the wind blowing because PowerPoint doesn't have a loop feature. It turns out though you can re-trigger an animation at the start of a sound file, and you can configure the sound file to loop. So you use that hack to make your animation repeat. That hack came from random googling, and didn't seem right. But when I bumped into some folks who help Werner with keynote content, I mentioned this as an aside and they said that yeah this is the way to do it, and they've used that trick plenty of times too.

Anyway, content wise this talk is pretty good. I had been giving a bunch of observability talks, where it walks you through logs, metrics, and traces, and alarms and stuff. But here I wanted to center the content on the outcome: fewer, shorter outages.