Hacker News Reader

A client of mine once told me "We are self-hosting our CI/CD servers because when problems occur, I can tell my employees to stay at work until it is fixed. I cannot tell GitHub to stay at work until they fix their issue."

View on HN

That seems like a weird way of measuring success (person-hours rather than some goal/outcome-oriented metrics like uptime).

I don't think that's what was intended. It reads to me there's more perception of control (employees directed to work) to achieve the goal (restoring service) during an outage. Outages happen regardless of uptime guarantees--it's just a matter of how often and how long they last.

That's how I understood it as well, and it's a pattern I've seen in (bad) decision making: While a stakeholder might feel more "in control" by being able to look over the shoulder of an on-call engineer, asking somebody to reboot the database, etc., this very often does not translate into higher performance metrics.

Of course, the opposite would be outsourcing all responsibilities, so there obviously is an art to balancing these two extremes.

Not only that, it's just odd logic. I'm going to prevent my employees from domain-specific value-add by forcing them to overwork on generic tooling? What?

Reply on news.ycombinator.com