A backup preserves state for restoration of a system. An archive preserves records for continuing retrieval and interpretation. The distinction is the purpose that preservation must satisfy, rather than the storage medium or file extension.
An application backup can contain every database row while requiring the original software to answer a historical question. An archive designed to replace that dependency must preserve the information and interpretation needed to answer the question through another access path.
Restoring operation and retrieving history
Restoration reconstructs a usable operating state. It depends on compatible software, configuration, keys, identities, and external services as well as copied data. The success criterion is the recovered system’s ability to resume required work.
Archival retrieval starts with a question about retained records. Which transaction occurred, what did a code mean at the time, which document supported it, or how was an amount calculated? The answer can require information that was distributed across tables, attachments, configuration, and executable rules.
A database export therefore supplies only the information it actually contains. A calculated report whose values were never stored needs either preserved calculation meaning or a preserved result appropriate to the historical purpose.
Integrity and interpretation
Integrity checks detect changes relative to a recorded reference. They do not establish that the reference was correct or that a reader can interpret it. The BagIt format described in RFC 8493 treats packaged payloads as opaque bytes even when completeness and manifest checksums validate successfully.
Take an intact file containing the values A, B, and C in a status column. If the code list has disappeared, the file can pass every checksum comparison while leaving the statuses uninterpretable. Preserving the bytes and preserving their meaning are separate deliverables.
Descriptive metadata supplies context for finding and understanding records. Schema definitions, code lists, relationship mappings, provenance, and descriptions of the retained population make the access path usable. Their exact content follows the historical questions the archive must answer.
Conversion can change the evidence
Format conversion changes representation. It can also change metadata and break references based on filenames, locations, or hashes. A converted document needs a traceable relationship to its source and to the records that refer to it.
The relevant question is not whether the new format opens. It is whether required content, associations, and interpretation survive the conversion. Attachments, external references, and historical identifiers expose failures that a successful file-open operation does not detect.
Where data is deliberately excluded or transformed, the archival record must explain that boundary. An apparently complete collection with unexplained omissions is harder to assess than a bounded collection with an explicit population and disposition.
Read-only is an access property
A read-only archive prevents specified users or processes from changing retained records through its access path. It does not by itself establish completeness, authenticity, retention compliance, or recoverability.
Authorized users still need to find the required information, while restricted users must remain unable to obtain records outside their permissions. Access controls and retrieval functionality therefore need separate evidence.
The archive also becomes an operated resource. Someone maintains its storage, catalog, access mechanism, and recovery. Removing the original application can simplify those responsibilities without eliminating them.
Acceptance follows the retained purpose
An archival acceptance exercise asks an authorized reader to answer representative historical questions without the live application. It includes difficult relationships and exceptions, rather than only the easiest search. An independent reader exposes interpretation that previously depended on undocumented operator knowledge.
The retained purpose and applicable obligations determine the population and duration. No universal legal retention period follows from calling a collection an archive. The fixed distinction is that backup success concerns restored operation, while archive success concerns usable retained information. A preservation design can serve both purposes only when it demonstrates both results.