@skevy@mjackson@spikebrehm The fallacy is thinking that you can create a service so small that you’ll never have to talk to another team and no external event will ever ruin your day.
@spikebrehm@mjackson I think the real nightmare is scaling development to thousands of engineers (but an important one, involving different tradeoffs than for small teams). We definitely made many mistakes and had a lot of chaos with microservices, but the status quo was also unstable.
Ruby’s still the language that I know best, and I have nothing but respect for how Rails completely changed the game on web dev (and it’s still a great way to get started quickly on a project). Other langs are more cost efficient and types make working on a large team better.
At @Airbnb, while we appreciated the flexibility of Ruby on Rails earlier on, it absolutely became unsuitable at scale (both traffic and organizationally) and we had to work miracles while replacing much of it with java microservices.
Gotta love the pivot from "it doesn't scale" to "you can't write real systems without types" as the new ding-dong argument against dynamically-typed languages like Ruby (and vanilla JavaScript and Smalltalk and...). https://t.co/lwxTGan6ZO