I was struck with downtime last week.
Linode’s Fremont data center went offline in a big way. Service was interrupted for nearly 4 hours.
Did I panic?
OF COURSE I DID!
At first at least. Thing is, when a site of mine goes down, I just assume that I had fucked something up.
Then I remember that I hadn’t pushed any code recently so it’s probably not me.
The next logical step is to check Linode’s status page which is followed by checking Twitter for being barking about their servers being down.
Twitter is usually up in arms faster than most companies can update their status page. The power of an angry consumer, amirite?
What always baffles me about this situation is how people want to put 100% of the burden on the hosting company in question.
Sure, their being down is on them, but the fact that you have a single server sitting in a single data center, that’s on you, bub.
Seems like everybody complaining had either a single server that they run everything on or they had multiple servers all in a single data center.
In Linode’s defense, only Fremont had issues that night. All of their other data centers were performing as expected.
Databases can be replicated across data centers. You can stand up web servers and load balance them as well.
Hell, you could have servers with completely different hosting companies if you really want to make things fault tolerant.
Point I’m trying to make here is that if you opt to put all of your eggs in one basket, you can’t be so pissed when things go awry.
If you really cared about your uptime, relying on a single point of failure is one hell of a way to show it.
Pretty sure there’s not a single hosting company on the planet that has been able to provide 100.00000% uptime, so it’s really only a matter of time before you experience any time of downtime.
High avilability is all about redundancy and having systems in place to [predictively] detect issues and mitigating them before they happen. Hard to pull that off with a single 5$ server with a virtual private hosting company.
With that, Linode’s outage put them at like ~99.5% uptime for the month, so egg on their face too ;)