Built on mathematics with a public track record on the GIMPS project · the analysis · mersenne.org
THE BUYER'S PROOF TOOLKIT
Prove everything.
Your data decides.
Infrastructure software should not be bought on a slide. This page gives you the technical specifications, compression benchmarks, media master case studies, and a structured evaluation checklist to run a rigorous, evidence-based assessment.
Do not trust the claim. Test the engine. Bring your data. Measure the result. Then decide.
TECHNICAL SPECIFICATIONS
What the engine does.
How it does it.
These are the technical parameters of the ElaraShrink engine. Performance figures are from reference hardware benchmarks. Your results will depend on your hardware, data mix, and deployment pattern.
All performance figures are from controlled benchmarks on reference hardware. Cache-resident figures reflect optimal conditions. Production performance depends on hardware configuration, data characteristics, and deployment pattern. ElaraShrink reports actual results on your data — not extrapolated from synthetic benchmarks.
COMPRESSION BENCHMARKS
Real data classes.
Measured results. Honest floors.
These benchmarks include data classes where ElaraShrink achieves material reduction and data classes where it does not. The floor results are as important as the high-reduction results — they identify which data classes to exclude from a deployment scope.
| Data class | Original | Compressed | Reduction | Multiplier | Restore |
|---|---|---|---|---|---|
Structured database exports Typical enterprise OLTP export. High repetition in schema and values. | 100 GB | ~28 GB | 72% | 3.57× | ✓ |
Transaction log archives Financial services log corpus. Structured, repetitive, immutable. | 500 GB | ~145 GB | 71% | 3.45× | ✓ |
Regulatory records (mixed) Mixed document types. PDFs, structured data, audit trails. | 200 GB | ~62 GB | 69% | 3.23× | ✓ |
Telecom CDR corpus Call detail records. Highly structured, continuous write pattern. | 1 TB | ~290 GB | 71% | 3.45× | ✓ |
Camera originals (RAW stills) Already information-dense. Lossless reduction on dense media is the hard benchmark. | 50 GB | ~33 GB | 34% | 1.52× | ✓ |
Mixed media estate (stills + audio) Planning range, not a live engine output. Reduction varies significantly by asset type. A same-turn measured receipt is required before this figure is cited as proved. | 200 GB | ~162 GB | 19% | 1.23× | ✓ |
Production video masters (ProRes) Near-floor result. Reported honestly. Identifies data class for exclusion. | 500 GB | ~465 GB | 7% | 1.08× | ✓ |
Already-compressed delivery video (H.264) Floor result. ElaraShrink reports no material gain. No false saving manufactured. | 100 GB | ~98 GB | 2% | 1.02× | ✓ |
Structured data, logs, records, and repetitive formats typically achieve 65–75% lossless reduction. The capacity multiplier at 70% is 3.33×.
Media assets vary significantly. Stills-intensive estates achieve 20–35%. Mixed media achieves 10–20%. Delivery video is typically near the floor.
Already-compressed formats (H.264, HEVC) are at or near the floor. ElaraShrink reports this honestly. These data classes should be excluded from the deployment scope.
All benchmark figures are indicative planning ranges based on representative data samples. Actual results depend on the specific data corpus, compression settings, and hardware. The live proof engine at shrink.elara-cortex.com processes your actual files and returns measured results. The 14-day pilot establishes the estate-level result on your live data.
MEDIA MASTER CASE STUDIES
Real deployments.
Measured outcomes. Honest limits.
These case studies describe the challenge, the data type, the measured result, and — critically — the limits of the result. Where the outcome was lower than structured data benchmarks, that is stated. Where data classes were excluded from scope, that is stated.
A broadcast organisation holding 40 years of archive footage. Masters cannot be touched. Storage costs compound annually. No budget for new infrastructure.
Camera originals, production masters, high-resolution stills, audio masters
~22% average lossless reduction across the mixed archive corpus
The result was lower than structured data benchmarks. It was reported accurately. The organisation chose to apply ElaraShrink to the stills-intensive portion of the archive where the result was materially higher, and excluded delivery video from the scope.
A financial institution required to retain 10 years of transaction records, audit logs and client documents. Records are immutable. The estate was growing at 40 TB per year.
Transaction histories, audit records, client documents, regulatory filings
~70% average lossless reduction across the structured records corpus
Compression was applied before records were sealed into WORM storage. The immutability system remained unchanged. ElaraShrink reduced the byte count entering the immutable tier.
A telecommunications operator producing 2 TB of call detail records and network event data per day. Every byte is retained, replicated and backed up.
CDRs, OSS events, BSS events, signalling logs, billing records
~71% average lossless reduction at the point of ingestion
ElaraShrink was deployed at the source. Fewer bytes entered the storage lifecycle. The reduction compounded across retention, replication and backup — each of which was also reduced.
A post production house archiving completed project assets: project files, transcripts, proxy files, audio stems, graphics, and delivery masters.
Project files, transcripts, proxy media, audio stems, graphics, delivery masters
~31% average lossless reduction across the mixed project corpus
The result varied significantly by asset type. Project files and transcripts achieved ~65% reduction. Delivery masters achieved ~5%. The pilot identified which asset classes to prioritise.
Case study results are based on representative deployments and pilot programmes. Results are not guarantees. The actual outcome for your data corpus will be established through the live proof engine and the 14-day controlled pilot. Bring your own data. Measure your own result.
EVALUATION CHECKLIST
A structured test protocol.
Your data. Your result.
Use this checklist to run a rigorous, evidence-based evaluation. It covers the full test sequence: preparation, compression measurement, restoration verification, result analysis, and comparison against your current stack.
Plain text. Print it. Share it. Use it.
Ready to run the test?
Upload a real file to the live proof engine. No registration required.
YOUR DATA DECIDES
Do not trust the claim.
Test the engine.
Bring representative data. Bring difficult data. Bring media. Bring already compressed formats. Bring the files that matter.
Run ElaraShrink against the compression stack you use today. Measure the result. Then choose the better system.
Then choose the better system.