Metacat's backend database stores the structure and data of any arbitrary XML document. The Metacat database uses the DOM approach to store XML documents as a hierarchy of nodes. All nodes are stored as records in database tables. The hierarchical relationship between the nodes is saved to preserve the structure of the documents.

The database Entity Relationship Diagram (ERD) currently looks like this:

• xml_documents contains document level information. It stands as a tie in to xml_nodes with the rootnodeid field.
• xml_nodes contains all of the data needed to recreate an XML document that has been inserted into the system. Each document's xml tree is represented by the node pointers in the table.
• xml_index contains indexing information for the paths in xml_nodes. This is data that has been denormalized for performance reasons. xml_index allows the fast searching of paths within an xml document.
• xml_revisions contains the document information (including the rootnode pointer) to any document that has been deleted (either by an explicit delete or by an update). This ensures that a complete lineage of any document is traced.
• xml_replication contains the list of Metacat servers that this server replicates to or from.
• xml_relation contains any relation information inserted by a Package document.
• xml_access contains all Access Control List (ACL) information relevent to a document.
• xml_catalog contains catalog information that represents DTDs and XSL stylesheets that are registered with the system.