2 Approaches on Uptime Calculation

1 min read
2-Approaches-on-Uptime-Calculation

100 percentWe’ve recently received questions about how do we calculate the uptime shown in our reports, so it seemed like a cool topic to show up here.

Uptime usually means the percentage of the time while the service was up, calculated by minutes. Things are usually clear when talking about past months or year, but various flavors    emerge when talking about current day, current month or current year.

1st approach

We calculate uptime by calculating the minutes of uptime divided by the total number of minutes for the specified period.

So, February has 29 days this year, 29 days x 24 hours x 60 minutes = 41760 minutes. 49 minutes of downtime, means that the site was up for 41711, and

41711 / 41760 = 0.9988, thus the 99.88% uptime. This means that if the service doesn’t have any more downtime this month, it would have a 99.88% uptime at the end of the month.

2nd approach

Another way to calculate the uptime is to take in consideration only the number of minutes already passed for current month and current year, which is perhaps what many of you have been expected to find. For February, this would have been:

7 days of February, 13 hours and 33 minutes = 10080 + 780 + 33 = 10893 minutes.

49 minutes of downtime would make it 10844, resulting in an uptime of 99.55%. The problem with this uptime is that it is only correct for one minute. Tomorrow, the uptime would increase as there were more minutes of February passed, making it: (12333 – 49) / 12333 = 99.60%.

So it would increase every day, until the end of the month when it would become 99.88% (the uptime currently been shown), or until another downtime would occur.

So basically, the uptime shown for the current month and year is actually the maximum uptime that the service could achieve this month/year. Whatever happens, that service will not have a greater than 99.88% uptime in February, or a 99.99% in 2012. For past years or month the uptime is calculated the same, so if in January you’d have a 49 minutes downtime, it would have a 99.88% uptime.

Conclusions

While the second approach is what you’d usually find in an uptime monitoring service, we’re currently going with the first approach, where data is more accurate and valid throughout the current time-span (current month, current year).

Interesting read?