The database administrators at site C have determined the following availability requirements. The table below provides examples.

Table 1. Sample moderate-to-high availability requirements
Question Answer Recovery technique
How many committed transactions can you afford to lose? A few minutes of transactions Enable after-imaging.
How long can the application be offline while you perform scheduled maintenance? Never Perform online backups with after-image files.
How long can you spend recovering if the database becomes unavailable? 24 hours Restore a full backup and a series of any number of incrementals or AI files.
Must distributed transactions occur consistently across databases? Yes Implement two-phase commit.

Given these requirements, the database administrators perform a full online backup every Saturday night. Because they can afford to lose only a few minutes of transactions, the site also enables after-imaging. They switch AI extents every day at 7 PM and archive the resulting full after-image extent. (Because they switch AI extents at scheduled intervals rather than waiting for an extent to fill, they must be sure to monitor AI extent usage in case an extent fills before the scheduled switch.) Because applications at this site perform distributed transactions that must be consistent across databases, this site implements two-phase commit. The full backup is tested Saturday night. After they are archived, AI files are applied to the tested backup since so few transactions can be lost.