Sites classified as highly available are designed to be ultra-resilient and offer the best protection from single failure points and underlying platform errors. The umbrella of availability incorporates failure, recovery and site resilience. Availability is commonly measured based on the time percentage that any system is working and active.
In order to improve availability it is necessary to put in place comprehensively engineered solutions. Since prediction of when or how a system will fail is notoriously difficult to predict, the best method of planning improved reliability is to engender shorter recovery times. If a given system is able to recover from a failure within 86.4 seconds then it’s possible to have a failure on a daily basis and still provide an availability of 99.9% to users.
