five nines calculation in uptime monitoring

Buyers and sellers of internet related services talk a lot about 5 9's uptime. Users often forget about the effect of the decimal place in a percentage representation in calculating allowable downtime for a particular service level agreement.

For example, since 1 percent is another way of expressing .01, then when speaking of five nine's uptime, or 99.999 percent uptime, then the related downtime calculation factor is not 0.001, but 0.01 of 0.001, or 0.00001, two more decimal places to the right.

The values have been precalculated in the following tables for yearly, monthly, and daily downtime totals for a range of values. This will help readers of the daily server uptime reports from the edgedirector.com managed dns services relate the percentage level of uptime shown in the report with actual downtime budgets. Other visitors who want similar reports can obtain free uptime reports from basicstate.com.

            

uptime       year            month              day  
             days  hh:mm:ss  days  hh:mm:ss     hh:mm:ss  

99.9990         0  00:05:15     0  00:00:27     00:00:01  
99.9900         0  00:52:34     0  00:04:28     00:00:09  
99.9000         0  08:45:36     0  00:44:38     00:01:26  
99.0000         3  15:36:00     0  07:26:24     00:14:24  

98.0000         7  07:12:00     0  14:52:48     00:28:48  
97.0000        10  22:48:00     0  22:19:12     00:43:12  
96.0000        14  14:24:00     1  05:45:36     00:57:36  
95.0000        18  06:00:00     1  13:12:00     01:12:00  
94.0000        21  21:36:00     1  20:38:24     01:26:24  
93.0000        25  13:12:00     2  04:04:48     01:40:48  
92.0000        29  04:48:00     2  11:31:12     01:55:12  
91.0000        32  20:24:00     2  18:57:36     02:09:36  
90.0000        36  12:00:00     3  02:24:00     02:24:00

One interesting approach to visualise the effect of multiple load balanced hosts on uptime is to consider two identical physical hosts. Presume that they are in two separate data centres on diverse networks. Presume further that each host can achieve 99.0 percent uptime and the outages are random. Then, all else being equal, for the 1.0 percent that host A is down, host B will be up 99.0 percent of those same times. Therefore, outages of the two hosts simultaneously is:

1.0 percent * 1.0 percent = 0.01 percent

or

0.01 * 0.01 = 0.0001

and, more rigorously

0.01 - (0.01 * 0.99) = 0.0001

Conversely, at least one physical host is available for:

100 percent - 0.01 percent = 99.99 percent

when two hosts with 99.0 percent uptime are deployed.

please link to this page instead of making copies