Abstract: An availability metric is obtained for groups of databases or other computer resources by capturing related stop and start events which are set. Event data, including the stop and start events, can be communicated from agents associated with the databases to a central manager. An availability monitoring server periodically queries the central manager to obtain the stop and start events, and analyzes the event data to determine time periods in which the databases are unavailable. The availability metric can be determined based on a ratio of: (a) a sum of the downtimes of the databases in a given time period and (b) a product of the total number of databases and the time period, e.g., expressed in units of database-minutes. The availability metric can be determined separately for separate clusters of databases and used to provide a service level agreement.