Failures and Recovery
In the previous chapter, you learned that failures are inevitable. Consequently, a well-designed and hardened application is all about early failure detection and quickly recovering from it.
Zero and 100 are very powerful numbers; no application can be 100% available, nor can it have a 0% failure rate. Only via great design and an appropriate level of tests can you ensure your application availability will tend toward 100%—but of course never reach it. Likewise, no matter how reliable your application is, it will fail at some point in time, and thus its failure rate is greater than 0%.
When your application is ...