The backup XML file contains a lot of paths to files, including the attachments, old cache, backup locations etc. The administrator can't view these without importing the file and if for example the attachments directory isn't where Jira expects it, the system is in inconsistent state when it's up after the restoration.

Optimally the backup process would work like this:

1) Tell Jira where the backup file is located
2) Jira scans the XML file and locates all the paths within the file
3) Administrator is presented with a page which lists those paths and allows editing them
4) Jira then restores the DB and replaces the paths stored in the XML with the paths the administrator provided