How To: Data Migration

The following is a general outline of ideas regarding our future data migration efforts:

Data migration to Omeka from other repository systems:

• Why?
  • We don't necessarily want to take the place of existing systems
  • We want to supplement and enhance existing systems
  • Keep existing system:
    • For legacy reasons
    • Too invested in existing system (time and money)
    • Dependent on a particular schema, controlled vocabularies, etc.
  • But use Omeka for its elegant and easy-to-use presentation layer
• How?
  • Omeka has a generalized schema that can accept almost any data model:
    • DCMI element set (Dublin Core) common to all resourses
    • Customized metadata (item types & metafields)
  • Straightforward data migration process:
    • Create parallel item types and metafields in Omeka
    • Map the existing database to the Omeka database
• Several ways to migrate data:
  • 1) Mediated import:
    • Existing database -> CSV/XLS/XML -> Omeka database
    • (+) Easy to code/implement due to the use of intermediary file formats
    • (+) Usually reusable across repository systems
    • (+) Can be done through Omeka's plugin architecture
    • (+-) Only as reliable as the export/mapping/import process
    • (-) Changes to existing database are not reflected in Omeka until another import is performed
  • 2) Un-mediated import:
    • Existing database -> Omeka database
    • (+-) A little more difficult to code/implement than mediated import
    • (-) Not usually reusable across repository systems
    • (+) Very reliable if implemented properly
    • (-) Changes to existing database are not reflected in Omeka until another import is performed
  • 3) ODBC (Open Database Connectivity):
    • Omeka request -> Existing database response -> Omeka system
    • (-) Difficult to implement (data mapping API must be perfect)
    • (-) Not usually reusable across repository systems
    • (+) Very reliable if implemented properly
    • (+) Changes to existing database are automatically reflected in Omeka
  • REST
  • OAI_PMH
  • etc.

(+) = positive attribute (-) = negative attribute (+-) = neutral attribute