My team works to keep the network uptime at Five 9's or better, which we interpret as 99.999% available. How do you track uptime?
How much downtime can you have in a year and still have Five 9's?
There are 31,536,000 seconds in a year,
Which is equal to 520,344 minutes,
Which is equal to 8,672.4 hours.
Let's work with minutes instead of hours or seconds. To keep the math simple, I'll round it to 500,000 minutes.
99% uptime is 495,000 minutes (500,000 minutes available - 495,000 actual = 5,000 minutes of downtime). Every time we move a decimal point, we can easily see the difference.
99.9% uptime will have only 500 minutes of down time.
99.99% uptime will have only 50 minutes of down time.
Five 9's--99.999% uptime means only 5 minutes of down time in a year.
How could you achieve Six 9's?
By playing games with semantics, and with setting exemptions that mean "certain outages don't count"? Usually those are planned outages. In that "politically adjusted world" the only things that count against your downtime count are "unplanned" outages.
Is that an accurate way to track it? Saying there were less than five minutes of unplanned outages means you hit your 99.999% target. But if your actual outages were in excess of 500 minutes, your uptime isn't even 99.9%. It still sounds pretty good, but if you were a client who needed to use the network and you experienced 500 minutes of service outage, you'd think it was not very good at all.
How does your organization track and count downtime? How can you achieve Six 9's?