Long-term stewardship · Field guide 08

Build a photo archive you can move again.

Every system is temporary. Portability is the ability to recover the files, meaning, relationships and evidence needed to keep the archive useful somewhere else.

Files alone are not an archive

A vendor may return every original file and still leave the agency unable to reconstruct the collection. Portability includes descriptive metadata, rights, relationships, identifiers, provenance, controlled vocabularies and enough documentation to interpret the export.

The Library of Congress identifies disclosure, adoption, transparency, self-documentation and external dependencies among the factors that affect long-term sustainability. Apply the same thinking to the archive as a system: prefer documented representations that can be inspected and processed with more than one tool.

Use stable identifiers

A path changes when storage changes. A filename may be reused. Assign an identifier that remains with the intellectual record across renames, derivatives and migrations. Preserve source identifiers too; they provide continuity with old catalogs, publications and records schedules.

Keep standard and local metadata legible

Use IPTC/XMP fields where their meanings fit. Define local fields when the agency has a genuine local need, then publish a data dictionary: name, purpose, type, cardinality, allowed values, sensitivity, source and export mapping. Avoid storing multiple concepts in one free-text field.

Components of a portable export
ComponentMinimum evidence
AssetsOriginals or resolvable references, identifiers, checksums, format information
MetadataDocumented CSV, JSON, XML or embedded standard fields with UTF-8 text
RelationshipsCollections, parents, derivatives, sequences and membership order
Rights and accessOwnership, usage terms, restrictions and sensitivity markings
ProvenanceSource, import batch, significant changes and review status
DocumentationSchema, vocabularies, export date, software version and known exceptions

Export while the system is healthy

Do not wait for a contract ending or application failure. Schedule metadata exports and periodic full packages. Keep the export specification under change control. When the schema changes, record the version and migration rule.

Calculate checksums before and after transfer. Store an inventory that associates each identifier with a file and checksum. A checksum demonstrates byte-level consistency; it does not prove that the file is correctly described or usable, so pair fixity checks with metadata and rendering tests.

Test a clean-room restoration

  1. Select a varied sample of assets and relationships.
  2. Export using documented production procedures.
  3. Give the package to someone without production access.
  4. Reconstruct records using only the export and documentation.
  5. Verify originals, fields, collections, rights and restricted values.
  6. Record gaps, revise the export and repeat.

An export is only trustworthy after someone has used it to reconstruct something meaningful.

Control external dependencies

Document any required codec, proprietary field, API, encryption key, identity service or cloud-only reference. Remove dependencies that prevent lawful backup and preservation. Where originals remain in agency-controlled storage, ensure the catalog export contains durable identifiers or paths that can be remapped.

Make portability a contract requirement

Specify export content, format, frequency, service levels, egress cost, professional-services needs and post-termination access. Require a sample export before award or early in implementation. Retest after major upgrades.

Sources