Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

If the reliability stays constant expressed as a rate (arguably optimistic), like ".1% failure per X flights", and the # of flights increases, then failures become more frequent.

Failure rates must decrease proportionally to the utilization/capacity increase for the # of failures to stay constant. Otherwise, the # of failures will naturally increase, it's simple multiplication.

For example:

If crime rates stay constant, but your population increases, you have more crime - not the same amount of crime as before. you have more terrorists, you have more school shootings, because your population doubled in size and you didn't cut the rates of abhorrent behaviors in half.

Air travel has _exploded_ in popularity, this is why you see all sorts of things more frequently than you did before. From reports of airline passengers being dragged off planes in totally asinine scenarios to deaths of pets in the cabin.



You're assuming the failure rate is related to the number of flights. With 2 failures in 20 years there's not really enough data to extrapolate.

And remember this wasn't a complete system outage - it affected 50% of flights, a degraded service.

In comparison the vast majority of startups that HN people work in will be lucky to last 2 years. In the last 10 years Gmail has been down, well not sure. It was down 6 times in one year. Uptime is apparently 99.8%, or down for 175 hours a decade, so 6 hours seems quite reliable.

Level 3 went down last month. And in 2016. And in 2015.

Having a system having to work for decades without downtime is unheard of in the web world, even in simple systems like gmail and AWS.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: