How getaccept achieved 100% uptime

I'm curious about how much time and effort was involved in building this environment.

Though I'd love to do it for my app, and I'm hoping and assuming this sort of environment will be available as a simple service in the near future, what was the decision which made it valuable enough to do this sort of engineering task as a start-up?

Netflix runs a similar environment, though they've also got the Simian Army regularly tearing around the joint.

If you're building a document signing service, and you are likely working between Europe and North America, you've got a perfectly acceptable update window, assuming you are able to roll an update out within a few minutes of downtime. You've spent X (hundreds?) of hours to save you those minutes, but will anybody notice?

Pushing a fix while on the phone with a customer may provide a wow factor, but would an update within a few hours not be equally wowable?

Possibly some insight into how many documents are signed per minute made this all necessary.

It's impressive, I'm admit, it just seems to fly in the face of all being lean as a start-up.