TDB2 - Database Administration

TDB2 directory layout

A TDB2 database is contained in a directory location DIR as:


where Data-NNNN are the compacted generations of the database. The highest number is the currently live database. The others are not used and not touched by the TDB2 subsystem. They can be deleted, moved elsewhere, or compressed as required. Each is a valid database in it own right.

Backups is the directory used to write backup files.

tdb.lock is the lock file to stop multiple use of the same database at the same time by different JVM processes. (If you wish to share a database between processes, or machines, consider using Fuseki2 with TDB2.


TDB2 databases grow over time as updates occur. They can be compacted by calling:


Compaction can be done on a live database. Read requests will continue to be serviced; write request are held up until compaction has finished. This can be a long time for large databases.

Compaction creates a new Data-NNNN subdirectory and copied over the latest view of the RDF dataset into that directory, then switch to using that generation of the database.

There is also a command line tool tdb2.tdbcompact to run the compaction process on a database not in use. The command line option --deleteOld removes the last database after compaction.

Compaction can also be called from the Fuseki HTTP Administration Protocol for live Fuseki webapps.


A TDB2 database can be backed up by calling:


which will create a dump file including a timestamp:


The file is a compressed N-Quads file.

Backup can be done on a live database. It takes a consistent view of the data and does not include any updates committed after the backup starts.

Backup can be called on a live database and read and write transactions continue to be serviced.

There is also a command line tool tdb2.tdbbackup to run the backup process on a database not in use.