We strongly recommend that you implement a regularly scheduled and frequent procedure for backing up your locations and data to avoid losses due to media, machine or operator failure. Backup rules for your installation will vary with each application. The frequency with which you run backups depends on the importance of the data, how often it changes, the maximum time needed to recover, and how large the data tables and location are. For each application, you must weigh the time and space costs of backups against the risk of data loss.
The BACKUP LOCATION and Restore batch files commands operate on a specific location and include all paths and files for all data objects in the location. In order to restore any combination of data objects from a particular location you must restore the entire location. The ROLLFORWARD command also operates on an entire location, applying a specified series of log files to a restored backup copy of the location data. See the following topics and the SQL Reference for detailed descriptions of the updated BACKUP LOCATION command and ROLLFORWARD command.
If you use an operating system backup command or another backup utility, be sure to back up all the files in all the paths that are defined for the location, including storage group (stogroup) paths and paths defined in various environment variables, if any. Be sure to archive and then delete the current series of log files after executing each location backup. Deleting the current log files resets location logging, causing a new series of log files to be built the next time the location is accessed.
If you want to use your operating system's backup, you must first shut down the server or take its locations off line, using the SQL OFFLINE command. See the SQL Reference for detailed descriptions of the ONLINE and OFFLINE commands.