Saving files to a cloud folder is not the same thing as building an archive. An archive must answer what was preserved, why the files matter, whether they remain unchanged, how they connect to their original context and whether the intended audience can still find and use them years later. The Gold Standard of Digital Archiving Services is a managed promise of trustworthy continuity—not a claim that unlimited storage alone solves the future.
Imagine a fictional secondary school preparing its fortieth-anniversary collection. Staff gather scanned newsletters, original digital photographs, old video recordings and interviews with teachers. A vendor imports everything into one portal and says the project is complete. Three years later, nobody can tell which photograph is from which year, an audio format will not open in current software and a private student interview appears in a public search. The files exist, but the archive is neither understandable nor safely governed.
The US National Archives and Records Administration’s digital preservation programme treats fixity, metadata, logged actions, validation and regular audits as essential. Its guidance was reviewed in March 2026. The Digital Preservation Coalition’s handbook similarly distinguishes preservation from ordinary file storage. These are professional frameworks—not US legal obligations automatically imposed on Singapore organisations.
Quick answer: what makes an archive trustworthy?
A trustworthy archive selects records for a clear purpose, preserves source files and provenance, checks integrity, records changes, controls access and demonstrates that people can still retrieve and understand the information. More cloud space does not automatically deliver these capabilities.
From the first file to a future reader
This chapter map follows selection, controlled ingest, preservation management, findability, privacy and long-term service exit.
1. Decide what deserves preservation
An organisation may have invoices, research outputs, videos, policy decisions, photographs, oral histories and countless temporary duplicates. It cannot preserve everything meaningfully simply by paying for more storage. Agree the collection’s purpose, appraisal criteria, responsibilities and any legal retention or disposal requirements before accepting files.
For the school’s anniversary, a signed founding letter and an interview with the first principal may merit lasting preservation. A temporary download cache may not. The selection rationale should be documented so future staff understand why a record was included or excluded.
Back to chapters · Next chapter
2. Separate archives, backups and records management
Backups primarily support recovery after loss or system failure. Records management controls creation, retention and disposition. An archive preserves selected material for ongoing evidential, historical or other value and sustains access to its context. A system may support more than one function, but the promises are not identical.
A daily backup that overwrites older copies does not provide a permanent trustworthy history. Conversely, a carefully preserved video archive may not restore tomorrow’s interrupted school administration system. Define the required service instead of purchasing ‘archive backup’ as an unexplained combined product.
Back to chapters · Next chapter
3. Recognise born-digital and digitised material as different sources
A photograph captured on a digital camera is born digital. A scan of a paper newspaper is a digital surrogate of a physical original. Both may need long-term care, but the scan’s meaning depends on its relationship to the source document. Record that relationship clearly.
If the school scans a handwritten staff letter, it should preserve relevant information about the original’s location, sequence and condition. The scan should not silently be presented as the physical original itself.
Back to chapters · Next chapter
4. Approve access rights before bringing material online
Archival selection does not create permission to publish. Photographs, interviews and documents can involve copyright, privacy, confidential information, contractual limits and restrictions imposed by donors. Determine who may access each class of record and what permission is needed for wider release.
A student’s private recollection of a difficult school experience should not become searchable worldwide merely because it was recorded for an internal oral-history project. Obtain and document relevant consent, restrictions and any later review decisions.
Back to chapters · Next chapter
5. Preserve the relationship between records and their context
A document without context may be impossible to interpret responsibly. Record the creator, originating department, period, project, version and relationship to other records where known. Do not use an arbitrary new filename as the sole description of provenance.
A photograph titled IMG_0427 has little retrieval value. ‘School science fair, 1994, photographer and rights unknown’ is more informative, while honestly documenting uncertainty rather than inventing names.
Back to chapters · Next chapter
6. Use an intake manifest that can be reconciled
At transfer, identify every file or logical object received, its original path or source, size, format and any existing checksum. Compare the manifest with the actual delivery before accepting the package as complete. The archive should distinguish files never received from files that subsequently became corrupted.
The Digital Preservation Coalition’s Getting Started guidance recommends file lists with format and checksum details, malware checks and spot-opening files. A vendor’s message saying ‘upload finished’ is not equivalent to verified ingest.
Back to chapters · Next chapter
7. Protect incoming devices and suspect files
Old external drives, memory cards, optical discs and shared network exports may be unreliable or contain malicious material. Establish an appropriate intake and isolation workflow before copying files into trusted systems. A production server should not be the first environment to open unknown content.
Where original media is fragile or unstable, involve appropriate digital-forensic or data-recovery specialists. The archive’s job is to preserve authenticity and data; it should not casually run destructive repair tools on the only original copy.
Back to chapters · Next chapter
8. Check content integrity with fixity information
A checksum or hash can help detect changes to file contents over time. Record a baseline at an appropriate point and repeat comparisons at controlled intervals. A mismatch needs investigation: corruption, authorised migration and unapproved alteration are different events.
The DPC handbook on fixity and checksums explains that checksums can support transfer, storage and retrieval verification. A matching checksum shows bit-level consistency with a reference; it does not independently prove a document’s authorship or the truth of every statement inside it.
Back to chapters · Next chapter
9. Capture descriptive, technical and preservation metadata
Good metadata answers different questions: what a record is about, its source, file format, dates, relationship to other objects, access conditions and what preservation actions were performed. Collect metadata sufficient for retrieval and future management without inventing information that was never known.
The US National Archives metadata requirements resource describes fields used in the transfer of permanent federal electronic records. That framework can inform local practice, but must not be called a mandatory Singapore archival schema.
Back to chapters · Next chapter
10. Keep original master files separate from access copies
Users may need small images, compressed audio or web-friendly documents, while preservation aims to retain significant properties and trustworthy source material. Maintain a controlled relationship between a preservation master and any derivatives supplied to the public. A thumbnail should never quietly replace the highest-quality surviving source.
NARA describes making separate public-use copies and tracking preservation actions. For the school, a web-sized picture can serve an exhibition while a carefully managed master supports future reprints and research.
Back to chapters · Next chapter
11. Plan formats for continuing usability
Files can remain intact while the software needed to open them disappears. Identify format types, documentation and dependencies such as fonts, codecs, databases or proprietary applications. Monitor which formats create a practical risk and develop a reasoned plan for transformation, emulation or another suitable preservation response.
The NARA Digital Preservation Framework discusses significant properties and format-specific risk plans. A conversion decision should preserve what matters to the record, not merely produce any file the current computer can open.
Back to chapters · Next chapter
12. Do not convert every object into one universal format
PDFs, photographs, email, software, moving images and complex scientific data can contain different important features. Flattening everything into screenshots may destroy search, structure, embedded data or interactivity. Choose preservation and access formats by record type and required properties.
A school science experiment spreadsheet contains formulas and data relationships that may matter beyond a static PDF. A performance video includes sound, motion and synchronisation. The archive’s selection of formats must be justified by those properties.
Back to chapters · Next chapter
13. Keep a dated audit trail of deliberate changes
Migration, restoration from backup, correction of metadata and creation of derivatives should leave a record showing who did what, when, why and which files were affected. Preserve relevant earlier information when making an authorised correction. Avoid overwriting the only copy and then asserting that history never changed.
The US National Archives says preservation actions are logged and files are audited. A service should show its own change-control process and be clear about what it can and cannot establish for a historical claim.
Back to chapters · Next chapter
14. Use independent copies and recovery testing
Storage devices, cloud accounts, providers and organisations fail. Preserve copies through a deliberate resilience design rather than assume one shared folder and its automatic sync are equivalent to independent protection. Consider separation of failure modes, controlled permissions and actual restoration tests.
The DPC handbook frames redundant storage, integrity monitoring and technology changes as ongoing work. A file existing in two folders on the same failing disk is not a meaningful continuity plan.
Back to chapters · Next chapter
15. Audit files after ingest, not just on opening day
Long-term preservation requires periodic checking of file existence, fixity, format readability, metadata and storage status. Define who owns the audit, how errors are investigated and how a known-good copy may be used to repair damage. Not every file needs manual opening every month, but no archive should rely on permanent silence.
NARA’s programme describes regular audits of preservation holdings and backup-restoration testing. A local supplier should propose a proportionate schedule rather than borrow NARA’s internal cadence and declare it a universal statutory requirement.
Back to chapters · Next chapter
16. Protect personal and confidential information
Historical files may contain children, staff, donors, medical details, residential addresses and confidential correspondence. Access controls, encryption where appropriate, retention review, incident management and disclosures should fit the information and Singapore’s applicable data-protection rules.
The Personal Data Protection Commission’s PDPA overview explains responsibilities involving personal data. An institution preserving a record for historical value still needs a defensible legal basis and appropriate access limits; ‘archive’ is not a universal privacy exemption.
Back to chapters · Next chapter
17. Give users a searchable and understandable catalogue
An archive becomes valuable when an authorised user can find records without memorising the original creator’s folder habits. Index titles, people and subjects where appropriate; allow sensible browsing; show dates, context, restrictions and known uncertainty. Test actual retrieval questions with representative users.
A former student might search for a 1990s science fair without knowing the name of the staff member who held the camera. Metadata and relationships should bring relevant items into view, while private student interviews remain appropriately restricted.
Back to chapters · Next chapter
18. Separate public access from ownership and authenticity
An item visible on a website may be a derivative, not the source. Show source context and relevant rights, and avoid presenting an unverified transcription as the authoritative original. Explain whether material is public-domain, licensed, permission-based or restricted.
NARA’s digitisation partnership principles distinguish ensuring digital-copy integrity from certifying historical authenticity. A portal should not turn the act of uploading into an evidentiary guarantee.
Back to chapters · Next chapter
19. Make accessibility part of the delivery design
An image-only scan can be difficult for someone using assistive technologies. Where practical and suitable, provide searchable text, readable structure, captions and alternatives, while distinguishing generated text from the original. Handwritten or damaged documents may require specialist transcription and quality control.
A generated transcription can help discovery but may contain mistakes. Keep it linked to the image source and identify uncertainty rather than rewriting the source to fit what software expected to read.
Back to chapters · Next chapter
20. Plan a vendor exit before signing a long contract
Ask how an archive can be exported with its objects, metadata, rights, relationships, checksums and audit history. The provider should define delivery formats, costs, access during migration and what happens when the subscription ends. An archive locked into a vendor’s private interface with no usable export is fragile.
Test a sample export early. A promise that ‘you can download everything’ is inadequate if filenames, metadata or links collapse into an unstructured pile on departure.
Back to chapters · Next chapter
21. Treat deletion and retention as controlled decisions
Not every record has indefinite retention rights. Define legal holds, retention review, authorised disposal and what happens to derived or backup copies where applicable. Document why material is retained or removed, and distinguish a restriction on public access from a decision to destroy the underlying record.
For the school, old data about individual minors may need tighter retention and access treatment than publicly distributed newsletters. The archival purpose should be defensible rather than assumed merely because the institution likes keeping files.
Back to chapters · Next chapter
22. Use a concrete retrieval test at acceptance
Before accepting the system, ask it to retrieve an agreed set of records by meaningful questions: a particular yearbook, one interview with restrictions, an archival-quality photograph and the source version of an old policy. Verify both content and permissions. Confirm that the export preserves the intended relationships.
If the anniversary archive can retrieve the science-fair photo but cannot say which event or year it belongs to, the metadata requirement has failed. If a private interview appears to a public user, the access design has failed even if the file is intact.
Back to chapters · Next chapter
23. Treat errors as preservation events
When a checksum changes unexpectedly, a vendor migrates files incorrectly or a wrong file becomes public, record and investigate what happened. Stop unsafe publication or destructive changes, identify which copies remain trustworthy and explain relevant impacts to the collection owner. Silence is not acceptable simply because the archive is expected to outlive its operators.
Correct the catalogue or file as authorised and leave an audit trail. Distinguish the technical repair from the historical claim, since a checksum cannot tell whether a photograph was labelled correctly in the first place.
Back to chapters · Next chapter
24. Measure future access rather than storage consumed
Terabytes stored and files uploaded are input measures. The stronger outcomes are verified completeness, integrity, appropriate metadata, discoverability, permission accuracy, successful restorations, understandable exports and files that remain usable after formats and staff change. Review the service through realistic future-reader tasks.
The gold standard is not a promise that technology will never change. It is a managed capability to preserve authenticity, detect damage, adapt carefully and keep knowledge accessible without losing its context or compromising the people represented in it.
The school’s fortieth-anniversary collection: a retrieval exercise
The school supplies 600 original photographs, 80 scanned newsletter pages and 20 recorded interviews. At ingest, the archivist records the source, manifest and integrity information, then identifies duplicates and missing context. Staff create descriptive records and separate public photographs from restricted interviews. The public exhibition uses access copies, while protected masters and audit logs remain under preservation management.
Three years later, a former student searches for a 1994 science competition. The archive finds the photographs and newsletter because their event, year and creator information were recorded. The source files still pass appropriate integrity checks. The former student cannot access a restricted interview because privacy rules are applied correctly. Those are three different demonstrations of success.
Digital archiving acceptance matrix
| Promise | Control the provider must maintain | Evidence an owner can test |
|---|---|---|
| Files accurately transferred | Manifest, validation and baseline checksums | Expected source items present and identified |
| History remains intelligible | Provenance, dates, relations and descriptive metadata | Meaningful search finds the right item |
| Files remain unchanged or changes are documented | Fixity, change logs and audits | Integrity report and investigated exceptions |
| Public access respects rights | Authorised role-based access and release decisions | Restricted records cannot be retrieved publicly |
| Archive can outlive its provider | Exportable masters, metadata, relationships and audit trail | Successful sample transfer and recovery |
Questions record owners often ask
Is a cloud drive a digital archive? Not by itself. Storage is a component; preservation also requires selection, context, integrity, rights, monitoring and sustainable access.
Does a file checksum prove a document is historically genuine? No. It can show bit-level consistency with a baseline but not that the original author or historical label was truthful. Provenance evidence is separate.
Does every old photograph need to be public? No. Copyright, privacy, consent and other restrictions may require limited access or withholding. Preservation and publication are different decisions.
Authoritative digital preservation sources and eduKate connections
Professional sources: NARA Digital Preservation Programme (reviewed March 2026); NARA Digital Preservation Framework; NARA Metadata Requirements; DPC Fixity and Checksums; Singapore PDPC Personal Data Protection Act.
Related eduKate articles: Libraries and Knowledge Access; Data Recovery Services; Document Shredding Services; Printing Services.