Configuring data-sources: Alternative IDs

Note: alternative IDs are only supported with Neo4j

When you save a visualization in Linkurious Enterprise, only the node and edge identifier are persisted in the user-data store, along with position and style information. When a visualization is loaded, the node and edge identifiers are used to reload the actual node and edge data from the graph database.

If you need to re-generate your graph database from scratch, the graph database will probably generate new identifiers for all nodes and edges, breaking all references to nodes and edges in existing visualizations.

Using a property as a stable identifier

You can configure Linkurious Enterprise to use a node or edge property as stable identifiers. Once set-up, Linkurious Enterprise will use the given property as identifier instead of using the identifiers generated by the database.

Thanks to this strategy, visualizations will be robust to graph re-generation.

Notice that the properties used as identifier should be indexed by the database to allow for a fast lookup by value.

Alternative identifiers configuration

To use alternative node and edge identifiers, edit your data-source database configuration in the configuration file (linkurious/data/config/production.json):

Example of alternative identifier configuration with Neo4j:

{
  "dataSources": [
    {
      "graphdb": {
        "vendor": "neo4j",
        "url": "http://127.0.0.1:7474/",
        "alternativeNodeId": "STABLE_NODE_PROPETY_NAME",
        "alternativeEdgeId": "STABLE_EDGE_PROPETY_NAME"
      }
      // [...]
    }
  ]
}

If you plan on using a compatible Neo4J version and alternative IDs, we recommend configuring alternative IDs indices.

To achieve better performance with alternative IDs is recommended to configure indices for alternative node and edge IDs.

Please note that the node or edge property used as an alternative ID must have string values. Other types (typically integers) are not supported. This is needed because the alternative IDs indices are FULLTEXT ones and can thus only index string values.

Refer to the documentation specific to Neo4j on how to configure these indices.