Thread by @RaeKnowler

Twitter was based on the idea of getting your friends together for a pizza or beer, and it ended up a platform for bad political ideas and harassing people.—@aspleenic #DevOpsDays

I got here a bit late today and just arrived in @aspleenic’s talk on reactive programming. A good point to jump in. :) #DevOpsDays

Problem with reactive programming: if you’re relying on following user feedback for building features, you find yourself always playing catchup. The users’ needs always outstrip your ability to add to your product. #DevOpsDays

Still, reactive programming is popular, because it’s better than months-long waterfall-style projects. E.g. Etsy deploys 25x a day. Wow, that’s fast! Way better than waiting 3 months for code to be deployed. #devopsdays

Okay, so, what’s proactive programming? (Nice 10th Doctor gif!) It’s about putting in time and effort to build the foundations of the project in advance. #DevOpsDays

Reactive programming: building a blanket fort. It’s quick! You can iterate.
Proactive programming: thinking about earthquakes and rain before you build a house, making sure it has a foundation and nice double-glazed windows. @aspleenic #DevOpsDays

People think PHP is insecure... nope! *Anything* is insecure if you don’t start with a focus on security. (Facebook was storing 90% of passwords in plaintext. 😭😭😭) #DevOpsDays @aspleenic

How do we start doing proactive programming? Incrementally (don’t just grab a buzzword and jump in!). You need stability & psychological safety: Things take longer at the start if you’re focusing on security and longevity. #DevOpsDays

Most important: a tight tight feedback loop! User feedback takes too long, that’s the end stage. #DevOpsDays

Tools are super important: for monitoring etc. (Humans are great at building tools! Like spoons, and hats.) @aspleenic #DevOpsDays

Chaos engineering! What happens if half your AWS instances go down at once? You need to know what will happen, how you can still provide the best service nonetheless. So try it!
1. You must never know when the chaos exercise is going to happen.
2. Do it in prod!
#DevOpsDays

Observability: we can’t just build a HTML site with a PHP form, put it in a server in a closet, and call it good any more. We use complex systems, and we need tools that give near-instantaneous feedback. #devopsdays @aspleenic

(We’re not usually building things that save lives! But we do want to build things that work and make people happy.)

If you’re logging into a server and tailing logs, you don’t have a full picture of what’s going on. Observability (with dashboards, graphs, etc) is key to an overview of your whole system... but if no one’s looking at it, all that data’s useless. @aspleenic #devopsdays

.@aspleenic’s not going to tell you to develop reactively or proactively. You need to choose what’s most important to your project, flexibility or stability. #devopsdays

@threadreaderapp Unroll, bitte schön.