7 comments

[ 4.4 ms ] story [ 27.0 ms ] thread
The article starts with "TL;DR: We believe...". I have to admit I stopped reading after the paragraph because I'm not really interested in what people believe, but evidence. It didn't help that it sounded like marketing-speak either.

Maybe I'm missing out on a great article, in that case I'll hope for other comments to change my mind.

The author seems to have a decent amount of experience using many services. He does breakdown each by price, memory, and any cons about them which is helfpul.
Thanks, the article was indeed interesting, even though reliability is what interests me the most. The "99.99% uptime SLA" mentioned on digitalocean's website means up to 3.5 days of downtime per year unless I'm mistaken.
Fair point, do you have any good resources that compares uptimes? I'd be curious to check that out.
99% uptime is 3.65 days (3 days 15hrs) of downtime.

99.9% uptime is 0.365 days (8 hrs 45 min) of downtime.

99.99% uptime is 0.0365 days (52 min 33 sec) of downtime.

It's also hard for the average person to catch unless they have users that are going to complain or a secondary service to constantly test the connection. And if they fail to deliver 99.99%, you just get a proportional credit. The benefits of advertising 99.99% usually outweigh the consequences of failing to deliver.

Aye, thanks. Shouldn't do math with a cold...

Edit: It also always depends on what uptime really means. I have one hoster who regularly does "network maintenance" where their entire network goes down for several hours. This can happen several times per year. So there's no easy failover.

It's so incredibly annoying. A few years back even their corporate site went down during such outages. They have that fixed now but I can't move all sites due to political reasons...