Yikes, just yikes... Quite a bit of misinformation in here.
I guess a bit more 'about me' is in order to qualify what I'm about to say
I have been involved in the hosting industry since 2002, in one form or another. From systems admin to support desk to (now) WHMCS, I've worn a ton of hats, including, at one point, working with an OnApp shop who shall remain nameless for a year and a half.
In case you're not familiar with it, Onapp is , it's only the largest cloud platform ever. I won't get into the competence of their support / staff / updates, because that's not the point here. The point is that OA is the beast to beat.
That said, on to the misinformation:
It seems that there's misconception that cloud means 100% uptime. This is absolutely, undeniably incorrect. While this can be close, it's not so much.
Case in point (something we saw a lot):
If the network your hypervisors are connected to is down, guess what? You're down.
The solution:
Multiple networks
Case in point
If the software solution you're using loads down the hypervisor? Guess what? You're down
The solution:
Load balancing hypervisors, making sure that the stripes are all correctly handled and off on other hv's
Case in point
If the hypervisor itself goes down, you're not terribly likely to get that backup working quickly.
Case in point
If the controller server goes down (say, for example OA staff logs in and screws things up, which they have a habit of doing), you're essentially screwed.
The point here? Cloud is not always 100%. yes, it may be close to it, but if you look at the infrastructure of most of these 'cloud' companies, you'd be surprised at how fallible, how vulnerable they really are. With most, all it would take would be one tiny little pindrop to set off a chain reaction so nasty, so ugly that it would be devastating.