This section refers to checking and backing up Linked Data View
and storage declarations only. The checks and backup/restore do not
affect physical quads, relational schema or tables or data therein.
For general backup and restore, see server administration. To
detect and fix automatically most popular sorts of RDF metadata
corruption use DB.DBA.RDF_AUDIT_METADATA . It is also
possible to backup RDF data by DB.DBA.RDF_BACKUP_METADATA and restore the
saved state later by using DB.DBA.RDF_RESTORE_METADATA . It is
convenient to make a backup before any modification of quad
storages, quad map patterns or IRI classes, especially during
debugging new Linked Data Views.
In SQL, adding a new view can not break anything. This is
because SQL lacks the ability of querying "everything" so data
sources are always specified. This is not true for SPARQL, so
please treat any metadata
manipulation as potentially destructive operation. If an RDF
storage is supposed to be used by more than one application then
these applications should be tested together, not one after other,
and they should be installed/upgraded on live database in the very
same order as they were installed/upgraded on instrumental machine
during testing. Always remember that these applications share RDF
tables so they may interfere.