Hacker News Reader

Comment by HtmlProgrammer | Hacker News Reader

Is this not just the general approach to system design? When I’m writing software, my first thought is list everything that could go wrong?

Or am I just a hopelessly anxious person lol

View on HN

I think it's the right approach, as long as you realize you won't be able to think of everything up front, and don't let that thought be a burden.

Simply being aware that there exist things that you don't know you don't know can save your project.

This is the general basis for why I tend to pick tools & concepts that are at least a half-decade old. The space of unknown unknowns in something that has been around this long should be vanishingly-small, especially if we are applying the tool or concept in a typical way.

> You have to think your optimization approach from the other end, as how it might fail.

But definitely give it some times to think of it upfront. Weight them by 2 factor: severity and frequency. Then try to tackle as much as you can from the top list of severity * frequency level.

Definitely agree it sounds like good system design. I think the overlap is the big picture thinking. It provides a way of re-framing goals to give a clearer picture of the most important things.

So not just list everything that could go wrong, but maybe: what's a terrible day for your service/system that's most likely to happen? Cascading failures? Outage that makes accessing/recovering your system impossible? Backups unusable?

Reply on news.ycombinator.com