Use cases

One hundred places
exact recovery decides.

Twenty-five enterprise domains, four evaluation scenarios each. Every scenario starts from the data a team already owns, names the measured workload class it belongs to, and states the boundary the result has to be judged against.

100 scenarios25 domains31 measured workload classes

What these are. Evaluation starting points drawn from the workload classes Saar actually measures. The evidence for each class is published at workload coverage.

What these are not. Not customer deployments, not endorsements, not certifications, and not measured savings. Exact recovery is the contract; the saving depends on the input and the baseline you compare against.

Saar is not encryption. Encrypted or near-entropy content may not shrink, and every result must be verified on representative files before acceptance.

Index

Twenty-five domains

Jump to the domain closest to your data. Each domain names the workload classes it draws on, so you can go straight to the measured sample.

Domain 01

Banking & payments

Payment rails, settlement records, regulatory evidence and cryptographic validation corpora. Retention outlives most of the systems that produced the data.

Workload classes: Blockchain & PQC data · Structured data · Email & messages · Network payloads

Keep NIST known-answer validation corpora intact for cryptographic release gates

Blockchain & PQC data

Cryptography lead · Platform security

Decision
Can the bank keep full conformance corpora for every algorithm version it ships, without the storage bill deciding what gets deleted?
Why it is hard
Known-answer test files are small in count but must remain byte-identical forever: a regenerated corpus proves nothing about the release that was actually shipped.
What Saar does
Store each corpus as a byte-exact archive with per-object recovery verification before acceptance.
Proof boundary
Verification is a byte comparison, not a hash of the codec output. Validate against your own release-gate procedure.

Archive settlement and statement records so an audit can replay the original file

Structured data

Head of payments operations · Internal audit

Decision
Which settlement exports move to cold storage, and can audit replay them exactly when a dispute reopens?
Why it is hard
Statements and settlement exports are regenerated from live systems that change; the archived file is the only record of what was actually reported.
What Saar does
Compress the retained exports under an exact-recovery contract, with the source digest recorded alongside each object.
Proof boundary
Confirm retention, legal hold and deletion obligations separately. Compression is not a records-management policy.

Retain signed payment message bundles without breaking signature checks

Email & messages

Payment scheme compliance · Messaging platform

Decision
How do we keep years of signed payment messages available for dispute evidence at a defensible cost?
Why it is hard
Message bundles are signed and sometimes encrypted; any transformation that alters bytes invalidates the evidence chain.
What Saar does
Compress the bundle losslessly and verify recovery per message before the archive is sealed.
Proof boundary
Signature validation must run against the restored bytes. Saar is not encryption and does not replace scheme controls.

Retain payment-rail packet captures for incident and fraud reconstruction

Network payloads

Fraud operations · Network security

Decision
Which packet captures survive the retention window, and can an investigation rely on them being unaltered?
Why it is hard
Captures are large, rarely read and expensive to keep, so teams shorten retention and lose reconstruction ability.
What Saar does
Compress captures for cold retention with exact recovery verified at write time.
Proof boundary
Replay tooling must accept the restored capture unchanged. Validate capture completeness at the source.

Domain 02

Insurance & actuarial

Policy wordings, claims evidence, actuarial models and rating runs. A claim can be reopened years after the file was archived.

Workload classes: Images · Text & documents · Structured data · Logs & observability

Keep claims evidence exact for reopened claims and regulatory review

Images

Claims operations director · Compliance

Decision
How much claims evidence can we retire, and what must stay exact for a claim that reopens in five years?
Why it is hard
Claims evidence mixes scanned documents, photographs and system exports; a reopened claim is judged against the file as it stood.
What Saar does
Archive evidence bundles with byte-exact recovery and a recorded source digest per claim file.
Proof boundary
Confirm the file is complete before archiving. Exact recovery does not establish authenticity or admissibility.

Retain every policy wording version exactly as issued

Text & documents

Product governance · Legal

Decision
Can we prove which wording was in force for a policy on a given date?
Why it is hard
Wordings are reissued constantly and downstream systems resolve only the current version; the issued document is the evidence.
What Saar does
Compress the issued-document archive losslessly, keeping version identity and issue date in the index.
Proof boundary
Index integrity must be validated separately from byte recovery. Compression changes neither wording nor effective date.

Archive actuarial model inputs and rating runs for reproducible reserving

Structured data

Chief actuary · Model governance

Decision
Can a reserving result be reproduced next year from the same inputs and configuration?
Why it is hard
Model inputs, parameter files and run logs are archived inconsistently, so reproduction depends on memory and re-export.
What Saar does
Archive the full run bundle as one exact unit: inputs, parameters, outputs and logs.
Proof boundary
Reproduction still requires the same model version and environment. Validate that separately.

Keep claims-system transaction logs for dispute and fraud timelines

Logs & observability

Claims systems owner · Internal audit

Decision
How long can transaction logs be retained at full fidelity, and can a timeline be reconstructed exactly?
Why it is hard
Logs grow fastest of any claims artefact and are usually sampled or discarded before a dispute needs them.
What Saar does
Compress log segments for long retention, verifying recovery per segment.
Proof boundary
Timeline logic depends on clock and correlation integrity at capture. Compression preserves bytes, not semantics.

Domain 03

Capital markets & trading

Order-book captures, FIX sessions, tick history and surveillance evidence. Replay has to reproduce the market as it was, byte for byte.

Workload classes: Structured data · Network payloads · Logs & observability

Preserve order-book captures for exact market replay

Structured data

Head of trading surveillance · Market data engineering

Decision
Which market-data captures stay available for replay, and will a replay reproduce the book exactly?
Why it is hard
Surveillance and best-execution reviews depend on replaying the market as it was, but captures are large and retention is usually cut first.
What Saar does
Compress capture files for retention with exact recovery verified before acceptance.
Proof boundary
Replay fidelity also depends on the feed handler. Validate capture completeness and sequence integrity.

Archive FIX session logs without losing the exact wire record

Network payloads

Electronic trading technology · Compliance

Decision
Do we have the exact wire-level record for a disputed order or fill?
Why it is hard
FIX sessions are logged at volume and often rotated away; reconstructing the wire record later is impossible.
What Saar does
Compress session logs losslessly for the retention window, keeping session identity in the index.
Proof boundary
Confirm clock and session integrity at capture. Compression does not repair an incomplete log.

Keep tick history reproducible for model research and backtesting

Structured data

Quant research lead · Data platform

Decision
Can a backtest be rerun from the exact tick data used the first time?
Why it is hard
Tick history is periodically re-cleaned and re-normalised, silently changing what a historical backtest would see.
What Saar does
Archive tick extracts as immutable exact objects, separate from the cleaned research copy.
Proof boundary
Backtest reproduction also requires the same code and parameters. Validate that separately.

Retain surveillance alerts and supporting data for the full investigation window

Logs & observability

Surveillance operations · Legal

Decision
Can an alert be re-examined with the same data years later, including the messages behind it?
Why it is hard
Alert evidence spans feeds, communications and reference data; retaining everything at full fidelity is treated as unaffordable.
What Saar does
Archive the alert evidence bundle with exact recovery per component.
Proof boundary
Reassessment depends on the analytics version. Compression preserves the evidence, not the model.

Domain 04

Digital assets & blockchain

Ledger snapshots, signed transactions, node state and proof artefacts. A restored byte that differs is a different transaction.

Workload classes: Blockchain & PQC data · Backups & archives · Indexes & transaction logs

Keep ledger snapshots restorable to the exact state a proof was made against

Blockchain & PQC data

Blockchain infrastructure lead · Security

Decision
Can we restore the exact ledger state behind a past attestation?
Why it is hard
Snapshots are large and quickly superseded; teams keep only the newest, so older proofs cannot be re-verified locally.
What Saar does
Compress snapshot archives with per-file recovery verification at write time.
Proof boundary
Chain verification must be rerun on restored data. Saar does not replace consensus or signature checks.

Archive signed transaction corpora for regression and compatibility testing

Blockchain & PQC data

Protocol engineering · QA

Decision
Which transaction vectors must remain available to test protocol changes against historical behaviour?
Why it is hard
Vector corpora accumulate and are pruned by storage pressure, weakening regression coverage exactly when a hard fork approaches.
What Saar does
Compress vector corpora losslessly and verify recovery before each release gate.
Proof boundary
Vectors must be validated by the client under test. Saar claims exact bytes, not behavioural correctness.

Retain node backups without inflating cold storage

Backups & archives

Node operations · SRE

Decision
How many node backup generations can be kept, and will any of them restore exactly?
Why it is hard
Node backups are large, mostly redundant and rarely tested; retention is traded against storage cost without recovery evidence.
What Saar does
Compress backup generations with per-object recovery checks recorded at write time.
Proof boundary
Restore drills remain mandatory. Compression is not a backup strategy.

Keep proof artefacts and witness data intact for later verification

Indexes & transaction logs

Cryptography engineering · Compliance

Decision
Can a proof produced last year still be verified from its original artefacts?
Why it is hard
Proof artefacts are small, numerous and often stored in scratch locations that are cleaned up after the release.
What Saar does
Archive proof bundles as exact objects with a manifest and per-file recovery check.
Proof boundary
Verification depends on the verifier version and parameters. Validate those separately.

Domain 05

Government & public records

Case files, census extracts, mail archives and statutory records. Retention schedules are measured in decades, not release cycles.

Workload classes: Text & documents · Structured data · Email & messages · Backups & archives

Keep case files complete and exact for statutory retention

Text & documents

Records manager · Programme director

Decision
Which case files can leave the active system while remaining producible on demand for the full retention period?
Why it is hard
Case files mix documents, correspondence and system exports; retention schedules are far longer than any single platform's life.
What Saar does
Archive case bundles with exact recovery verified per file and a manifest per case.
Proof boundary
Redaction, access control and disposal rules stay in the records system. Compression is transport, not policy.

Retain census and survey extracts reproducibly for statistical reanalysis

Structured data

Chief statistician · Data governance

Decision
Can a published statistic be recomputed from the extract that produced it?
Why it is hard
Extracts are re-derived and re-cleaned, so the dataset behind a published table may no longer exist in its original form.
What Saar does
Archive published extracts as immutable exact objects with a release identifier.
Proof boundary
Disclosure control must be applied before archiving. Compression does not anonymise data.

Retain official mail archives without breaking authenticity checks

Email & messages

IT records lead · Information governance

Decision
How do we keep official correspondence available and verifiable for the retention period?
Why it is hard
Mail archives are large, frequently migrated and lose signature validity when converted between stores.
What Saar does
Compress mailbox exports losslessly, verifying recovery per mailbox before the migration is signed off.
Proof boundary
Signature validation must run against restored messages. Compression does not re-sign or repair headers.

Store departmental archival exports at a defensible cost

Backups & archives

Departmental CIO · Archive service

Decision
Can departments hand over large archival exports without a storage budget fight?
Why it is hard
Archival exports are rarely accessed but must remain recoverable for decades, so cost drives premature disposal.
What Saar does
Compress transfers as sealed archives with recovery verification at handover.
Proof boundary
Format obsolescence is a separate risk. Exact recovery does not guarantee future readability.

Domain 06

Defence & national security

Cryptographic validation corpora, firmware bundles, packet captures and execution-state evidence on disconnected estates.

Workload classes: Blockchain & PQC data · Firmware & binary · Network payloads · Memory & execution states

Hold post-quantum test vectors on disconnected estates

Blockchain & PQC data

Crypto authority · Secure systems engineering

Decision
Can every air-gapped site hold the full validation vector set it needs to certify a build?
Why it is hard
Disconnected sites cannot pull vectors on demand, so they keep local copies that are costly to store and awkward to update.
What Saar does
Compress vector sets into sealed bundles with per-file recovery checks before distribution.
Proof boundary
Certification remains with the authorised lab and process. Compression is not an approval.

Archive signed firmware releases for the lifetime of the platform

Firmware & binary

Platform integration lead · Supply chain security

Decision
Can a firmware release from a decade ago still be reproduced and verified from its original bundle?
Why it is hard
Signed images are archived in multiple stores with inconsistent naming, and older releases are pruned as storage grows.
What Saar does
Compress release bundles losslessly with a signature-preserving manifest and recovery proof per image.
Proof boundary
Boot-chain and signature validation must be rerun. Compression does not alter or replace signing.

Retain mission network captures for post-mission analysis

Network payloads

Mission systems lead · Cyber operations

Decision
Which captures survive the deployment, and can analysis be repeated on the original traffic?
Why it is hard
Deployed storage is constrained, so captures are downsampled or discarded before the analysis team sees them.
What Saar does
Compress captures under an exact-recovery contract, verifying per capture before it leaves the platform.
Proof boundary
Classification handling and release rules are separate. Compression does not change the classification.

Preserve memory images for forensic review without altering them

Memory & execution states

Digital forensics lead · Incident response

Decision
Can a memory image be stored for later review and still be treated as unaltered evidence?
Why it is hard
Memory images are large, captured under time pressure and often reduced to extracts that cannot answer later questions.
What Saar does
Compress the image for retention with exact recovery verified at capture time and a recorded digest of the source.
Proof boundary
Chain of custody and tool validation remain procedural requirements. Compression is not evidence handling.

Domain 07

Healthcare providers

Imaging volumes, clinical messages, audit trails and correspondence. Clinical and legal review can reach back years.

Workload classes: Medical imaging volumes · Structured data · Logs & observability · Text & documents

Retain imaging volumes for longitudinal review at sustainable cost

Medical imaging volumes

Chief medical information officer · Imaging director

Decision
Which imaging studies move to long-term archive, and can a radiologist review the original volume years later?
Why it is hard
Volumes are large and multiply with every follow-up; archiving is expensive, so older studies are compressed or discarded without recovery proof.
What Saar does
Compress study archives with per-series recovery checks at ingest.
Proof boundary
Diagnostic interpretation requires validated viewers and clinical governance. Compression is not a diagnostic claim.

Keep clinical messages exact for care audit and continuity

Structured data

Clinical informatics lead · Patient safety

Decision
Can we show exactly what a clinical message said when care was delivered?
Why it is hard
Interface engines rewrite and transform messages, so the stored version may not be the version the clinician saw.
What Saar does
Archive delivered messages as immutable exact objects with delivery identifiers.
Proof boundary
Clinical interpretation requires context and clinician judgement. Compression changes nothing about the message.

Retain audit trails for access review and investigation

Logs & observability

Privacy officer · Security operations

Decision
How long can access audit trails be kept at full fidelity, and can an investigation reconstruct them?
Why it is hard
Audit trails grow fastest of any clinical system artefact and are routinely abbreviated before an investigation needs them.
What Saar does
Compress audit segments for retention with per-segment recovery verification.
Proof boundary
Identity resolution and clock integrity remain prerequisites. Compression preserves bytes, not identities.

Keep clinical correspondence and referrals producible

Text & documents

Health records manager · Legal

Decision
Can correspondence and referral letters be produced exactly when a complaint or claim arrives?
Why it is hard
Correspondence is scattered across mailboxes, portals and scanned files with inconsistent retention.
What Saar does
Compress correspondence exports losslessly, verifying recovery per batch.
Proof boundary
Redaction and consent rules apply before archiving. Compression is not a redaction control.

Domain 08

Life sciences & genomics

Sequencing output, instrument captures, structure files and analysis artefacts. Reanalysis must start from the original read.

Workload classes: Genomic & biological data · Scientific simulation outputs · Structured data · Images

Retain raw sequencing reads for reanalysis

Genomic & biological data

Genomics core lead · Research data steward

Decision
Which raw reads stay available for reanalysis, and can a pipeline restart from the original read?
Why it is hard
Raw reads are large and usually superseded by processed outputs, yet reanalysis with a new pipeline needs the original input.
What Saar does
Compress raw read archives with exact recovery verified per sample.
Proof boundary
Pipeline validity depends on reference and tool versions. Compression preserves the read, not the biology.

Keep instrument captures exact for method reproducibility

Scientific simulation outputs

Lab operations lead · Quality assurance

Decision
Can a measurement be reproduced from the original instrument output?
Why it is hard
Instrument software exports change across versions, and archived outputs are often re-exported rather than kept.
What Saar does
Archive raw captures as sealed exact objects with acquisition metadata.
Proof boundary
Instrument calibration and method parameters remain part of reproduction. Compression does not record method state.

Retain structure and assay datasets for model retraining

Structured data

Computational biology lead · Data science

Decision
Can a model be retrained on exactly the dataset that produced the published result?
Why it is hard
Training sets are assembled from many sources and quietly change as sources update.
What Saar does
Archive training snapshots as immutable exact objects with a dataset identifier.
Proof boundary
Licensing and consent constrain reuse. Compression does not change data rights.

Store microscopy and pathology captures for later review

Images

Pathology director · Research imaging

Decision
Which image captures remain reviewable years after the study closed?
Why it is hard
Whole-slide and high-content captures are large, and review copies replace originals in practice.
What Saar does
Compress capture archives with per-image recovery verification.
Proof boundary
Image interpretation is a clinical and scientific judgement. Compression is byte preservation, not image quality.

Domain 09

Medical devices & imaging

Voxel volumes, device telemetry, firmware images and service logs across equipment lifetimes measured in decades.

Workload classes: Medical imaging volumes · Firmware & binary · Telemetry & signals · Logs & observability

Retain device-produced voxel volumes across the equipment lifetime

Medical imaging volumes

Biomedical engineering lead · Clinical engineering

Decision
Can a scan produced by a retiring device still be reviewed exactly?
Why it is hard
Device-specific formats are often readable only by the original vendor software, so the archive must be preserved as captured.
What Saar does
Compress volumes as exact objects with recovery verification at export.
Proof boundary
Vendor software and format support are separate risks. Validate viewer availability before disposal.

Keep device firmware history for safety investigations

Firmware & binary

Quality and regulatory affairs · Device security

Decision
Can we produce the exact firmware that was running at the time of an incident?
Why it is hard
Firmware updates overwrite prior versions on the device, and archives are inconsistent across model families.
What Saar does
Archive each released image with a signature-preserving manifest and per-image recovery check.
Proof boundary
Regulatory reporting and device verification remain separate obligations.

Retain device telemetry for reliability and post-market surveillance

Telemetry & signals

Post-market surveillance lead · Reliability engineering

Decision
How much device telemetry can be kept, and can a failure signature be re-examined exactly?
Why it is hard
Fleet telemetry is high volume and usually aggregated, discarding the raw signal needed to diagnose a rare failure.
What Saar does
Compress raw telemetry segments for retention with per-segment recovery verification.
Proof boundary
Diagnosis also depends on device metadata. Validate capture completeness at the gateway.

Keep service and maintenance logs producible for audits

Logs & observability

Field service operations · Compliance

Decision
Can a maintenance history be produced exactly for an audit or warranty dispute?
Why it is hard
Service records live in multiple systems and are often summarised, losing the original engineer notes and attachments.
What Saar does
Compress service record exports losslessly with per-record recovery checks.
Proof boundary
Record completeness is a workflow control. Compression preserves what was recorded.

Domain 10

Energy & utilities

Grid telemetry, meter archives, protection logs and control-system state. Regulators and insurers ask for the original record.

Workload classes: Telemetry & signals · Structured data · Logs & observability · Memory & execution states

Retain grid telemetry for regulatory and insurance review

Telemetry & signals

Grid operations manager · Regulatory affairs

Decision
How long can raw grid telemetry be kept, and can an event be reconstructed exactly?
Why it is hard
Telemetry volumes are enormous and typically aggregated within days, losing the resolution an event investigation needs.
What Saar does
Compress telemetry segments for retention, verifying recovery per segment before acceptance.
Proof boundary
Clock synchronisation and sensor calibration remain prerequisites. Compression preserves samples, not meaning.

Archive meter and billing data for dispute and settlement

Structured data

Revenue assurance lead · Billing operations

Decision
Can a disputed bill be settled from the original meter record?
Why it is hard
Meter data is rolled up into billing runs, and the raw interval data may be gone by the time a dispute arrives.
What Saar does
Compress interval archives with exact recovery per meter and period.
Proof boundary
Tariff and estimation logic are separate from data preservation. Validate those in billing.

Keep protection and control logs exact for incident review

Logs & observability

Protection and control engineer · OT security

Decision
Can a protection operation be reviewed from the exact relay and control logs?
Why it is hard
OT logs are often exported to flat files and truncated, and the sequence matters more than the summary.
What Saar does
Compress log exports losslessly for the review window with per-file recovery checks.
Proof boundary
Relay configuration and firmware context are required for interpretation. Compression preserves bytes only.

Snapshot control-system state for change and incident forensics

Memory & execution states

Control systems lead · Change management

Decision
Can we restore the exact control-system state that existed before a change?
Why it is hard
State is captured inconsistently across vendors, and snapshots are large enough that only the latest is retained.
What Saar does
Compress state snapshots with exact recovery verified at capture.
Proof boundary
Restoration into live OT requires vendor procedures and safety review. Compression is not a rollback tool.

Domain 11

Oil, gas & mining

Seismic surveys, drill records, geospatial models and simulation output. Survey value outlives the acquisition campaign.

Workload classes: Seismic data · Text & documents · Geospatial & environmental data · Scientific simulation outputs

Keep seismic surveys available for reprocessing

Seismic data

Subsurface data manager · Exploration lead

Decision
Which surveys stay available for reprocessing, and can a reprocessing run start from the original field data?
Why it is hard
Survey archives are among the largest assets a company holds; retention decisions are made on cost and often destroy reprocessing optionality.
What Saar does
Compress survey volumes with per-line recovery verification at archive time.
Proof boundary
Reprocessing depends on geometry and acquisition metadata. Validate those separately.

Preserve drilling and well records for regulatory and partner review

Text & documents

Well operations lead · Regulatory compliance

Decision
Can a well history be produced exactly for a regulator or joint-venture partner?
Why it is hard
Well records span decades, contractors and formats, and summarised reports replace originals in practice.
What Saar does
Compress well record bundles losslessly with per-document recovery checks.
Proof boundary
Regulatory format requirements are separate. Compression does not change a record's authority.

Retain geospatial and reservoir models for revisit

Geospatial & environmental data

Geoscience data lead · Asset team

Decision
Can a reservoir model be reopened years later exactly as it was delivered?
Why it is hard
Models are tied to software versions and are often migrated between formats, with loss.
What Saar does
Archive model packages as sealed exact objects with a manifest and recovery proof.
Proof boundary
Software version compatibility is a separate risk. Validate before decommissioning tooling.

Keep simulation runs reproducible for field decisions

Scientific simulation outputs

Reservoir engineering lead · Technical assurance

Decision
Can a development decision be re-justified from the original simulation inputs and outputs?
Why it is hard
Runs are archived partially, so the decisive case may not be reproducible when a decision is revisited.
What Saar does
Archive complete run bundles, with exact recovery per file.
Proof boundary
Reproduction needs the same simulator version. Validate separately from byte recovery.

Domain 12

Manufacturing & industrial

CAD and BIM models, MES records, inspection imagery and line telemetry. A recall or warranty dispute reopens the archive.

Workload classes: Industrial & CAD designs · Structured data · Images · Telemetry & signals

Retain CAD and BIM models for the life of the asset

Industrial & CAD designs

Engineering data manager · Plant engineering

Decision
Can the as-built model be produced exactly when a modification or insurance review needs it?
Why it is hard
CAD suites and formats change faster than plants; migrated models lose fidelity quietly.
What Saar does
Archive model packages as exact objects with per-file recovery verification.
Proof boundary
Modeling standards and approvals are separate. Compression preserves what was delivered.

Keep production and quality records producible for recall traceability

Structured data

Quality manager · Manufacturing systems

Decision
Can we show exactly what the line produced and recorded for a suspect batch?
Why it is hard
MES records are summarised into quality reports, and the raw production records are aged out.
What Saar does
Compress production record exports with per-batch recovery checks.
Proof boundary
Traceability also requires lot and serial linkage. Validate that in MES.

Retain inspection imagery for warranty and safety review

Images

Inspection lead · Warranty operations

Decision
Can an inspection image be re-examined exactly years after the part shipped?
Why it is hard
Inspection images are numerous and often compressed or downscaled at capture, losing the detail a later review needs.
What Saar does
Archive original captures with exact recovery verified at ingest.
Proof boundary
Capture resolution and lighting are fixed at source. Compression cannot recover lost detail.

Keep line telemetry for downtime and quality analysis

Telemetry & signals

Operations engineering · Continuous improvement

Decision
Can a quality excursion be analysed from the original sensor stream?
Why it is hard
Historian aggregation hides the short excursions that explain a defect, and raw segments are not retained.
What Saar does
Compress raw telemetry segments for review windows with per-segment recovery checks.
Proof boundary
Sensor calibration and tag mapping remain part of interpretation.

Domain 13

Automotive & mobility

LiDAR and SLAM maps, CAN traces, ADAS logs and firmware. Reconstruction after an incident depends on exact sensor bytes.

Workload classes: Automotive & robotics mapping · Telemetry & signals · Firmware & binary · Memory & execution states

Retain LiDAR and SLAM maps for validation and incident review

Automotive & robotics mapping

Autonomous systems lead · Validation engineering

Decision
Which map and sensor captures stay available, and can a validation run be reproduced from them?
Why it is hard
Mapping captures are enormous and usually retained only for the current release, while incidents can surface months later.
What Saar does
Compress capture sets with per-file recovery verification at ingest.
Proof boundary
Localisation quality depends on calibration and processing. Compression preserves raw captures only.

Keep CAN and vehicle bus traces exact for diagnostics

Telemetry & signals

Vehicle diagnostics lead · Field quality

Decision
Can a field failure be diagnosed from the original bus trace?
Why it is hard
Traces are captured ad hoc and exported to reduced formats that drop arbitration detail.
What Saar does
Compress trace archives losslessly with per-file recovery checks.
Proof boundary
Diagnosis also needs vehicle configuration and software level.

Archive ECU software versions for recall and update traceability

Firmware & binary

Software release manager · Homologation

Decision
Can we produce the exact software that was flashed into a vehicle?
Why it is hard
Update campaigns move fast and older binaries are pruned, while recall evidence depends on the exact image.
What Saar does
Store signed images with signature-preserving manifests and per-image recovery proof.
Proof boundary
Flashing and signature checks stay in the vehicle pipeline. Compression does not affect signing.

Capture hardware-in-the-loop state for test reproduction

Memory & execution states

Test automation lead · Validation

Decision
Can a failing HIL scenario be replayed exactly after the bench has moved on?
Why it is hard
Bench state is ephemeral and captured inconsistently, so a failure may not reproduce.
What Saar does
Archive state captures with exact recovery verified per session.
Proof boundary
Replay depends on bench configuration and models. Validate that separately.

Domain 14

Aerospace & space

Flight telemetry, ground-station captures, mission science products and avionics bundles. Downlink budgets and certification both bite.

Workload classes: Telemetry & signals · Network payloads · Scientific simulation outputs · Firmware & binary

Retain flight-test telemetry for certification evidence

Telemetry & signals

Flight test engineer · Certification authority liaison

Decision
Can a certification point be re-derived from the original telemetry?
Why it is hard
Test campaigns produce enormous telemetry, and retention is often reduced to analysed parameters.
What Saar does
Compress raw telemetry with per-channel recovery verification at archive time.
Proof boundary
Certification evidence also requires calibration and traceability records.

Archive ground-station captures for anomaly investigation

Network payloads

Mission operations lead · Ground segment

Decision
Can an anomaly be investigated from the exact downlink capture?
Why it is hard
Downlink windows are short and captures are stored in ad hoc formats before being deleted to free capacity.
What Saar does
Compress capture archives with exact recovery checks per pass.
Proof boundary
Frame decoding depends on mission configuration. Validate against the ground segment.

Retain mission science products for long-term research

Scientific simulation outputs

Science data centre lead · Principal investigator

Decision
Which science products remain available for reprocessing by the research community?
Why it is hard
Products are versioned by pipeline, and older versions are retired while papers still cite them.
What Saar does
Archive product versions as immutable exact objects with recovery proof.
Proof boundary
Calibration and pipeline versions must be recorded alongside the data.

Keep avionics load images for airworthiness review

Firmware & binary

Airworthiness engineer · Software quality

Decision
Can we produce the exact loadable software that was approved?
Why it is hard
Load images are managed in toolchains that change, and approved artefacts are not always retained as delivered.
What Saar does
Archive load images with signature manifests and per-image recovery checks.
Proof boundary
Airworthiness approval remains with the authority. Compression is not a compliance artefact.

Domain 15

Telecommunications

Call-detail records, signalling captures, element logs and subscriber archives. Lawful-access and billing disputes need originals.

Workload classes: Network payloads · Logs & observability · Graph databases

Retain call-detail records for billing and lawful-access obligations

Network payloads

Mediation and billing lead · Regulatory compliance

Decision
How long can raw call-detail records be kept, and can a billing dispute be settled from them?
Why it is hard
Records are enormous and typically aggregated into charging events, discarding the raw record a dispute needs.
What Saar does
Compress raw record archives with per-file recovery verification.
Proof boundary
Lawful-access authorities and processes are separate. Compression is not a retention mandate.

Keep signalling captures for network and fraud investigation

Network payloads

Network security lead · Fraud management

Decision
Can a signalling incident be reconstructed from the exact capture?
Why it is hard
Signalling systems rotate captures rapidly, and the sequence needed for analysis is gone within hours.
What Saar does
Compress captures for the investigation window with exact recovery checks.
Proof boundary
Protocol decoding depends on the stack version. Validate with your own tooling.

Retain element logs across vendor upgrades

Logs & observability

Network operations centre · Vendor management

Decision
Can an outage be reviewed from logs produced before a vendor upgrade?
Why it is hard
Logs change format between releases and are usually retained for days, not the incident-review window.
What Saar does
Compress log segments losslessly with per-segment recovery verification.
Proof boundary
Log parsing depends on release-specific schemas. Preserve schemas alongside the log.

Keep network topology graphs for outage and capacity analysis

Graph databases

Network planning lead · OSS engineering

Decision
Can an outage or capacity decision be analysed against the topology as it stood?
Why it is hard
Topology is stored as current state in OSS tools, and historic graph dumps are rarely retained.
What Saar does
Compress periodic topology exports with exact recovery per snapshot.
Proof boundary
Discovery completeness affects the graph. Validate coverage against OSS records.

Domain 16

Cloud & data centres

Container layers, object exports, VM disks, configuration bundles and platform logs. Storage and egress costs scale with bytes.

Workload classes: Cloud & object storage · Backups & archives · Servers & infrastructure · Virtualization artifacts

Store container layer inventories and images with exact restore

Cloud & object storage

Platform engineering lead · Registry operations

Decision
Which image generations stay pullable, and can a rebuild be traced to the exact layers?
Why it is hard
Registry storage grows without bound, and garbage collection silently removes the layers an old deployment pinned.
What Saar does
Compress layer archives with per-layer recovery verification before retention.
Proof boundary
Registry integrity and signing remain separate controls. Compression does not verify provenance.

Retain object-storage exports at a lower cost without losing fidelity

Backups & archives

Cloud storage owner · FinOps

Decision
Which exports leave the hot tier, and will any restore be byte-identical?
Why it is hard
Colder storage tiers promise savings but teams cannot prove restores are exact, so they either overpay or over-delete.
What Saar does
Compress exports with recovery verification per object and a manifest.
Proof boundary
Cost comparisons must include compute, egress and restore time, not bytes alone.

Keep configuration bundles exact for drift and incident review

Servers & infrastructure

Infrastructure lead · Change management

Decision
Can the exact configuration that was running during an incident be recovered?
Why it is hard
Configuration is stored as rendered state, and the bundle actually deployed is overwritten by the next release.
What Saar does
Compress configuration bundles losslessly with per-bundle recovery checks.
Proof boundary
Secrets must be handled separately. Do not archive credentials into a general store.

Retain VM disk snapshots for rollback and migration

Virtualization artifacts

Virtualisation lead · Platform operations

Decision
Which VM snapshots stay recoverable, and will a restored disk match the snapshot exactly?
Why it is hard
Snapshots multiply quickly and are pruned by age, so the generation needed for a rollback or migration may already be gone.
What Saar does
Compress snapshot archives with per-disk recovery verification at write time.
Proof boundary
Restores must be drilled on the target hypervisor. Compression is not a snapshot mechanism.

Domain 17

Software & SaaS

Repositories, build artefacts, model weights, customer backups and incident forensics. Reproducibility depends on the exact archive.

Workload classes: Source code · Backups & archives · AI models & embeddings · Logs & observability

Keep repository mirrors restorable for build reproducibility

Source code

Engineering productivity lead · Release engineering

Decision
Can a release from two years ago be rebuilt from the exact sources used?
Why it is hard
Mirrors are pruned and history is rewritten, so old builds cannot be reproduced from current state.
What Saar does
Compress repository snapshots with exact recovery per snapshot.
Proof boundary
Reproducible builds also depend on toolchains and pinned dependencies.

Retain customer backup generations with proven restores

Backups & archives

SaaS operations lead · Customer success

Decision
How many backup generations can we promise, and can we prove each one restores?
Why it is hard
Backups are retained by policy but rarely restore-tested, so the promise is untested.
What Saar does
Compress backup generations with per-object recovery checks recorded at write time.
Proof boundary
Restore drills and tenant isolation remain mandatory. Compression is not a backup product.

Archive model weights and embeddings for rollback and audit

AI models & embeddings

ML platform lead · Model governance

Decision
Can we roll back to the exact model that produced a past decision?
Why it is hard
Weights are large and overwritten in place; the model behind an audit sample may be gone.
What Saar does
Compress weight archives with per-file recovery verification.
Proof boundary
Determinism also depends on runtime and hardware. Compression preserves weights only.

Preserve application state and logs for incident forensics

Logs & observability

Site reliability lead · Security

Decision
Can an incident be reconstructed from the state and logs as they were?
Why it is hard
Containment destroys state, and log retention is shorter than the review cycle.
What Saar does
Compress forensic bundles with exact recovery per artefact.
Proof boundary
PII and secret handling apply before archiving. Compression does not sanitise.

Domain 18

Retail & ecommerce

Catalogue and POS exports, product imagery, clickstream logs and supplier documents. Peak-season replay and audit both need originals.

Workload classes: Structured data · Images · Logs & observability · Text & documents

Retain catalogue and price snapshots for dispute and audit

Structured data

Pricing operations lead · Commercial finance

Decision
Can we prove the price and description a customer saw on a given day?
Why it is hard
Catalogue data is overwritten continuously, and dispute evidence is reconstructed from screenshots.
What Saar does
Compress periodic catalogue snapshots with exact recovery per snapshot.
Proof boundary
Channel and tax rules affect what the customer saw. Capture those alongside the snapshot.

Keep product imagery masters for every channel

Images

Digital asset manager · Brand operations

Decision
Which image masters stay available when a channel needs a re-render?
Why it is hard
Masters are replaced by derived crops, and re-delivery requires the original.
What Saar does
Compress image masters with per-asset recovery verification.
Proof boundary
Colour management and rights are separate. Compression does not change licensing.

Retain clickstream and funnel logs for attribution disputes

Logs & observability

Digital analytics lead · Marketing finance

Decision
Can an attribution decision be recomputed from the original events?
Why it is hard
Pipelines aggregate early, and attribution disputes arrive after the raw events have expired.
What Saar does
Compress raw event segments for the dispute window with per-segment recovery checks.
Proof boundary
Consent and data-protection limits apply. Compression does not create lawful basis.

Keep supplier and compliance documents producible

Text & documents

Supplier compliance lead · Quality

Decision
Can a supplier certificate or declaration be produced exactly as received?
Why it is hard
Documents are stored across email, portals and shared drives, often converted and degraded.
What Saar does
Compress document sets losslessly with per-document recovery checks.
Proof boundary
Certification validity is a compliance judgement. Compression does not validate a certificate.

Domain 19

Media & entertainment

Delivery masters, audio stems, game builds and archive libraries. Masters are retained for decades and re-delivered repeatedly.

Workload classes: Video · Voice & audio · Haptic & telepresence data · Game & app assets

Retain delivery masters for re-versioning and licensing

Video

Post-production lead · Rights management

Decision
Which masters stay online, and can a re-delivery use the exact approved master?
Why it is hard
Masters are large and duplicated across facilities, and the approved version can be hard to identify later.
What Saar does
Compress masters with per-file recovery verification and a manifest of approved versions.
Proof boundary
Codec and container conformance are delivery requirements. Validate separately.

Keep audio stems and session archives for remix and audit

Voice & audio

Audio post lead · Music licensing

Decision
Can a mix be revisited exactly from the stems used?
Why it is hard
Sessions reference external samples and plugins, and stems are archived in inconsistent formats.
What Saar does
Compress stem and session archives with exact recovery per file.
Proof boundary
Plugin and DAW versions affect reproduction. Archive those notes with the session.

Preserve motion-capture and haptic session data for re-render

Haptic & telepresence data

VFX supervisor · Animation pipeline

Decision
Can a shot be re-rendered or re-targeted from the original capture?
Why it is hard
Capture sessions are large and tied to pipeline versions, and retargeting needs the raw motion data rather than the baked animation.
What Saar does
Compress capture sessions with exact recovery per take.
Proof boundary
Rig and solver versions affect reproduction. Record them alongside the capture.

Retain build and patch payloads for support and certification

Game & app assets

Build engineering lead · Platform certification

Decision
Can we reproduce the exact build a player is running?
Why it is hard
Builds are superseded quickly and platform archives retain only the latest certified build.
What Saar does
Compress build payloads with per-file recovery checks and build identifiers.
Proof boundary
Platform certification and signing are separate. Compression preserves payload bytes.

Domain 21

Education & research

Course archives, research data, assessment records and institutional repositories. Long retention with small budgets.

Workload classes: Text & documents · Scientific simulation outputs · Backups & archives · Video

Retain course materials and assessment records for appeals

Text & documents

Registry director · Academic quality

Decision
Can an assessment decision be reviewed against the materials students actually received?
Why it is hard
Learning platforms version content in place, so the module a student studied may no longer exist.
What Saar does
Compress course packages with per-file recovery verification at each release.
Proof boundary
Assessment policy and marking remain academic governance matters.

Keep research data available for replication

Scientific simulation outputs

Research data steward · Library services

Decision
Can a published finding be replicated from the deposited data?
Why it is hard
Deposits are often transformed for repositories, and the analysis-ready version is lost.
What Saar does
Compress deposited datasets losslessly with exact recovery per file.
Proof boundary
Ethics and consent constrain reuse. Compression does not grant access rights.

Store institutional repository content at sustainable cost

Backups & archives

Scholarly communications librarian · IT

Decision
Can the repository keep everything it has accepted without repeated storage crises?
Why it is hard
Repository growth outpaces budget, and format migration is deferred until files are at risk.
What Saar does
Compress repository objects with recovery verification and fixity checks.
Proof boundary
Format migration remains a preservation activity. Compression is not migration.

Retain exam and proctoring captures for integrity review

Video

Examinations officer · Academic integrity

Decision
Can an integrity case be reviewed from the original capture?
Why it is hard
Proctoring captures are large and deleted quickly after results are released.
What Saar does
Compress capture archives with exact recovery for the review window.
Proof boundary
Privacy and retention limits apply. Compression does not extend lawful retention.

Domain 22

Scientific computing & HPC

Simulation output, instrument data, sky surveys and climate grids. Petabyte archives are reanalysed for years after the run.

Workload classes: Scientific simulation outputs · Geospatial & environmental data · Astronomical data

Keep simulation checkpoints for restart and verification

Scientific simulation outputs

HPC operations lead · Computational science

Decision
Which checkpoints stay available, and can a run be restarted exactly?
Why it is hard
Checkpoint storage dominates facility budgets, so old checkpoints are deleted before results are verified.
What Saar does
Compress checkpoints with per-file recovery verification at write time.
Proof boundary
Restart fidelity also depends on solver version and decomposition.

Retain climate and weather grids for reanalysis

Geospatial & environmental data

Climate data lead · Modelling group

Decision
Can a reanalysis start from the exact model output that was produced?
Why it is hard
Grids are remapped and compressed at production, so the native output is no longer available.
What Saar does
Compress native grid archives with per-file recovery checks.
Proof boundary
Grid metadata and calibration remain part of interpretation.

Preserve survey images and catalogues for follow-up

Astronomical data

Survey data manager · Observatory

Decision
Can a detection be re-measured from the original survey frame?
Why it is hard
Frames are processed into catalogues and the raw frames are archived selectively.
What Saar does
Compress frame archives with exact recovery per exposure.
Proof boundary
Photometric calibration is part of the science pipeline, not the archive.

Retain detector events for reanalysis

Scientific simulation outputs

Computing coordinator · Physics experiment

Decision
Can a published measurement be re-derived from the original events?
Why it is hard
Event data is reduced by trigger and reconstruction, and the raw sample cannot be re-derived once discarded.
What Saar does
Compress event archives with exact recovery verification per block.
Proof boundary
Detector conditions and calibration are required for physics interpretation.

Domain 23

Semiconductors & EDA

GDSII and OASIS layouts, mask data, verification artefacts and test programmes. A tape-out archive is a legal and engineering record.

Workload classes: IC design & semiconductors · Industrial & CAD designs · Structured data

Archive GDSII layouts for tape-out and legal record

IC design & semiconductors

Physical design lead · CAD engineering

Decision
Can the exact layout that was taped out be produced years later?
Why it is hard
Layouts are large, tool-version dependent and often migrated between PDKs, with silent differences.
What Saar does
Compress layout archives with per-file recovery verification and a recorded design revision.
Proof boundary
Design rule and PDK versions must be recorded alongside the layout.

Retain mask and OASIS data for fab and audit

IC design & semiconductors

Mask operations lead · Fab engineering

Decision
Can a mask set be re-fabricated from the original data?
Why it is hard
Mask data is transferred to fabs under strict controls, and retained copies are inconsistent.
What Saar does
Compress mask data with exact recovery and transfer manifests.
Proof boundary
Export control and confidentiality obligations apply. Compression does not change them.

Keep verification artefacts for regression and sign-off evidence

Industrial & CAD designs

Verification lead · Design quality

Decision
Can a sign-off result be re-examined from the artefacts that produced it?
Why it is hard
Coverage and wave databases are huge and pruned to save space, weakening later analysis.
What Saar does
Compress verification artefacts with per-file recovery checks.
Proof boundary
Tool versions and testbenches are required for reproduction.

Retain test programmes and results for yield analysis

Structured data

Test engineering lead · Yield analysis

Decision
Can a yield excursion be traced to the exact test programme and results?
Why it is hard
Test programmes change with every revision and results are aggregated before analysis.
What Saar does
Compress programme and result archives with per-file recovery verification.
Proof boundary
ATE configuration and calibration are part of interpretation.

Domain 24

Logistics & maritime

EDI interchange, shipment telemetry, route captures and customs documentation. Disputes are settled from the original message.

Workload classes: Network payloads · Telemetry & signals · Geospatial & environmental data · Text & documents

Preserve EDI interchange for dispute and settlement

Network payloads

EDI operations lead · Commercial finance

Decision
Can a disputed shipment be settled from the exact message exchanged?
Why it is hard
EDI messages are processed and archived inconsistently across partners and value-added networks.
What Saar does
Compress interchange archives losslessly with per-message recovery checks.
Proof boundary
Trading-partner agreements govern message authority. Compression does not alter them.

Retain fleet telemetry for safety and insurance review

Telemetry & signals

Fleet safety manager · Insurance liaison

Decision
Can an incident be reviewed from the original vehicle telemetry?
Why it is hard
Fleet platforms aggregate data quickly, and incident review happens weeks later.
What Saar does
Compress raw telemetry segments for the review window with recovery checks.
Proof boundary
Device calibration and clock integrity remain prerequisites.

Keep route and mapping captures for network planning

Geospatial & environmental data

Network planning lead · Operations analytics

Decision
Can a route decision be revisited with the map state at the time?
Why it is hard
Maps are updated continuously, and historic route performance is evaluated against today's network.
What Saar does
Compress map and route capture archives with exact recovery per release.
Proof boundary
Attribution of delay to route requires operational context.

Keep customs documentation exactly as declared

Text & documents

Customs compliance lead · Trade operations

Decision
Can a declaration be produced exactly as filed?
Why it is hard
Declarations are stored in broker systems, and the filed version is often replaced by amendments.
What Saar does
Compress declaration archives with per-document recovery verification.
Proof boundary
Legal validity of a declaration is determined by the customs authority.

Domain 25

Agriculture & environment

Remote sensing, soil and yield records, weather grids and sensor streams. Multi-season comparison needs the original capture.

Workload classes: Geospatial & environmental data · Structured data · Telemetry & signals

Retain remote-sensing captures for multi-season comparison

Geospatial & environmental data

Agronomy data lead · Precision agriculture

Decision
Can a field's condition be compared across seasons from the original captures?
Why it is hard
Imagery is processed into indices, and the original captures are discarded before multi-year comparison.
What Saar does
Compress capture archives with exact recovery per scene.
Proof boundary
Atmospheric correction and sensor calibration are part of the comparison.

Preserve soil and yield records for certification and insurance

Structured data

Farm operations manager · Certification body liaison

Decision
Can a certification or claim be supported from the original records?
Why it is hard
Records are held in farm-management platforms that change, and exports lose structure.
What Saar does
Compress record exports losslessly with per-file recovery checks.
Proof boundary
Certification decisions rest with the certifier.

Keep weather grids for agronomic model reproduction

Geospatial & environmental data

Modelling lead · Advisory services

Decision
Can an advisory recommendation be reproduced from the weather inputs used?
Why it is hard
Weather feeds are reprocessed continuously, so the grid behind an advisory may no longer exist.
What Saar does
Compress grid archives with exact recovery per run.
Proof boundary
Model version and parameters are part of reproduction.

Retain field sensor streams for irrigation and compliance review

Telemetry & signals

Irrigation manager · Environmental compliance

Decision
Can a water-use or irrigation decision be reviewed from the original sensor stream?
Why it is hard
Sensor data is summarised at the gateway, and regulators ask for the raw series.
What Saar does
Compress sensor segments with per-segment recovery verification.
Proof boundary
Sensor calibration and placement affect interpretation.

Coverage

Measured classes behind the catalogue

31 of the 31 measured workload classes are represented above. Each class has a named real sample, a recorded comparison and a stated limitation.

Open workload coverage

Plan an evaluation on your own files

Back to the index