Migrate your database

You can migrate databases between different versions of Kuzu without manually writing DDL and COPY statements. To perform a migration, you would first export the database from an existing Kuzu version using the command EXPORT DATABASE. Then, you would import this database into another Kuzu version using IMPORT DATABASE.

Export database

The EXPORT DATABASE command exports all the contents of a Kuzu database to the specified directory. The command generates the following files under the /path/to/export directory, which are needed to import this database to a new Kuzu version:

• schema.cypher: Contains the definitions of all node and edge tables.
• macro.cypher: Includes the definitions of all macro functions.
• copy.cypher: Contains COPY FROM statements for reimporting data.
• Data files: These files contain the actual database data (in CSV or Parquet format).

The parameters for EXPORT DATABASE are the same as those used when running COPY TO statements. The example below shows how to export a database using default configurations.

EXPORT DATABASE '/path/to/export'

The default behaviour is to export the data tables to Parquet format. This is done to reduce the scope of formatting errors and to improve performance. You can use the format parameter to output to other formats such as CSV. The format parameter can either be csv or parquet.

Here’s an example of how to export a database using the CSV format for data files.

EXPORT DATABASE '/path/to/export' (format="csv", header=true);

Note

The EXPORT DATABASE command only exports indexes whose dependent extensions have been loaded.

Import database

The IMPORT DATABASE command imports the contents of the database from a specific directory to which a Kuzu database was exported using the EXPORT DATABASE command. Under the hood, this command uses the Cypher and data files that were created by the EXPORT DATABASE command to recreate the necessary databases and tables in the new Kuzu version you are migrating to.

The query below imports the database from a directory named /path/to/export to the current database that your session points to:

IMPORT DATABASE '/path/to/export';

Note

1. The IMPORT DATABASE command can only be executed on an empty database. Currently, in case of a failure during the execution of the IMPORT DATABASE command, automatic rollback is not supported. Therefore, if the IMPORT DATABASE command fails, you will need to delete the database you are connected to and reload it.
2. The IMPORT DATABASE command also imports all indexes, regardless of whether their dependent extensions were loaded during import. If the dependent extension was not loaded during import, it will be automatically loaded during import.

Export database with schema only

Similar to SQLite’s .schema command, you can use the schema_only = true option to export only the database schema without exporting the data itself:

EXPORT DATABASE '/path/to/export' (schema_only=true);

This command will generate a schema.cypher file containing only the DDL (Data Definition Language) statements that can be used to recreate the schema of the database inside Kuzu.