The discussion of downtime is, unfortunately, subjective. Most Cloud (hosted) providers will give uptime assurances of between 99.9% (three nines) and 99.999% (five nines). But what does that mean? Well, for starters, anything less than three nines is a service level agreement you should walk away from. Why? Well, take a look at the monthly breakdown:
[table id=1 /]
Extrapolating from this data, an SLA of three nines means your service can be down for 8.75 hours per year. Five nines is 5.25 minutes of downtime annually. The more nines, the higher the service guarantee.
That’s stunning, to me. 99.9% uptime never seemed so bad. But it gives your provider the go-ahead to 10 unscheduled minutes of downtime per week.
What’s subjective about these hard numbers, though, is how your vendor defines downtime. There’s no industry standard. Is it a total service outage? does intermittent connectivity count as uptime? What if there’s a delay, but not an outage? What if I can “sort of” work?
I put all my marbles in Cloud solutions, but realize no SLA is ironclad. And, regretfully, I feel better about this when I see major corporations go down (lookin’ at you, United).
So what’s the moral of the story? that there are no 100% guarantees, and if you’re promised 100% uptime you should consider ocean front property in Arizona. But, if you can better understand what these nines actually promise you, and then define an outage, you can make an informed about any Cloud service.