W hen facing a natural catastrophe or a failure of a local system, Google's basic principle to ensure constant availability of its services is always the same: redundancy. Data is systematically replicated in multiple data centers located in different areas. Moreover, robust failover software mechanisms ensure service continuity, whatever happens.

Another important architecture principle at Google, which helps ensure optimal availability, is a thorough decoupling of software from hardware. For example, no p rocess depends on the availability of a particular piece of hardware for its execution.

To inform its customer in a transparent way, Google provides a web site, called the Apps Status Dashboard, as shown in the previous screenshot, which lists, grouped by date, each malfunction that occurred on any of the Google Apps. It also indicates the actions undertaken by Google teams to fix them and it gives an estimate for the time necessary to restore the situation to normal.