Hacker News Reader

100% uptime is impossible and the vast majority of companies do not need anywhere near the 99.99% they claim is necessary, nor can they actually pay for it.

A 30-second outage to switch to a replica is perfectly fine for production.

View on HN

If you have a replica, you have two db's already as I've suggested. And no one said anything about uptime; the point was single point of failure, not uptime. When I said single master, I just meant a single db as opposed to a master/slave setup.

The GP said single master, not single database.

I am the GP, I think I know what I meant better than you, which is why I clarified that in the comment you just replied to.

Reply on news.ycombinator.com