Background: Know: availability, reliability, outage, Availability Calculation, Recognize: Measuring availability in 9s
Up | ||
Previous | ||
Risks in Availability %ages[]
This page discusses the potential risks in measuring availability over ages. It shows that both reliability and availability should be considered.
Risks
If the availability is measured over a long period, it may be difficult to differentiate a few long outages from many short outages. However, users are more sensitive of long outages than short outages because they can not wait for a long time to access the service. Therefore, they may avoid the service or even turn to other providers. In addition, this will infulence the credibility and lose more market share.
Measurement Taken
The service performance should be monitored and Over exceedingly long interval outages should be measured. In addition, finer-grained, ongoing metrics of such things as system availability should be deployed.
Examples
In 2001, most of Microsoft major web sites could not be accessed for 24 hours because of an operation error. This had a bad influence on the credibility of the company. Moreover, if Microsoft wanted to achieve "five nines" availability, it had to operate outage-free for the next 250 years, which was almost impossible.
Calculation: 250 years * 365 days/year * (1-0.99999) = 0.9125 =approx= 1 day
See also
Reference
1: ↑ Tim Moors, 2012, TELE9752 Network Operations and Control - Fault Management, pp. 12-15.