According to a dictionary, uptime means:
the period of time when a computer, system, etc.
is being used without any problems.
What that means in real time:
| Uptime | Total downtime per year |
|---|---|
| 99.00000% | 3 days, 15 hours and 39 minutes |
| 99.90000% | 8 hours, 45 minutes and 56 seconds |
| 99.99000% | 52 minutes and 35 seconds |
| 99.99900% | 5 minutes and 15 seconds |
| 99.99990% | 31 seconds |
| 99.99999% | 3 seconds |
(source: Matt Rickard – @mattrickard at Twitter)
So this is downtime per year!
At Fuga you get a 99.99% uptime SLA,
although we are actually between 99.99% and 99.999%. But let’s assume 2×9 behind the dot:
52 minutes and 35 seconds per year
is 3155 seconds per year
Divide by 365 days makes (rounded up) 8.64383 seconds per day
x7 = 60.50681 seconds per week
That’s an itsy bitsy teenie weenie more than 1 minute of downtime per week!
(source: Matt Rickard – @mattrickard at Twitter) For every extra 9 after the dot you should take into account a multiplication of 10 for the costs. If you know what it will cost you to be down for 1 minute, you can make an offset against the costs you will incur for wanting that extra 9 behind the dot in your SLA. Because everything is possible, we know that, but these kinds of considerations deserve a fair calculation.
