Patent 02 / Selective restore architecture
01 / 12US10671315B2
↑↓  navigate  ·  all patents →
Siten Sanghvi / Granted June 2, 2020

Restore what matters.
Not everything.

A blockchain architecture for selectively copying and reconstructing production data in lower environments, without repeatedly cloning the entire system.

US10671315B2Patent number
2018-08-17Filed
1 yr 9 moTime to grant
18 Claims3 independent
19 CitationsForward citations
SCROLL TO TRACE THE SYSTEM
Visual patent explainer
01 / The operational pain

A full clone is a blunt instrument.

Large enterprises copy production environments for testing, quality assurance, and development. Traditional cloning copies the whole database, even when a team only needs a relevant slice.

The patent says this can consume days or weeks and make environment refreshes so expensive that they may happen only once or twice a year.

Traditional production copy
01
Clone the entire systemProduction data, code, records, and unrelated history all travel together.
OVER-COPY
02
Wait for the restoreNetwork, storage, and compute resources remain occupied during a lengthy migration.
SLOW
03
Test on a stale copyExpensive refresh cycles leave lower environments behind production.
STALE
02 / The inventive move

Treat history as a selectable ledger.

Store system data on a blockchain, classify the data inside blocks, and selectively clone only the section required by the target environment.

This is not merely a blockchain backup. The controller reconstructs a purposeful subset, can append later increments, and can combine multiple selected sections into a useful lower environment.

Parent blockchain / selective reconstruction
BLOCK 01AccountsH:a431
BLOCK 02OrdersH:bb72
BLOCK 03ConfigH:c091
BLOCK 04ArchiveH:d173
BLOCK 05LogsH:e805
TARGETRebuild01 + 02 + 03
selectedsource chaintarget environment
03 / Data flow

From production to a purpose-built copy.

A controller connects to the source chain, chooses a section, clones it, and reconstructs that section in a target system used for QA, testing, or development.

The patent also describes nodes, an entity system, system data storage, and a user device that initiates or manages the restore.

Simplified architecture
01 / SOURCE ENVIRONMENTProduction system data stored as sequential blockchain blocks.
02 / RESTORE CONTROLLERConnect, qualify, select, copy, and reconstruct.
03 / END BLOCK RULESSet cutoffs per data type. Copy only the required history.
04 / TARGET ENVIRONMENTQA, testing, development, or a combined lower environment.
04 / The data anatomy

Blocks know what they contain.

The architecture buckets data by type and generates separate hashes for those types. The block header stores the separate hashes, so the restore process can qualify and select data with precision.

That modified header is central: it turns the chain into a navigable structure for selective reconstruction, not just an immutable sequence.

Modified block header / typed integrity
ACCOUNTH: 4ac1e8
ORDERH: 7bf283
CONFIGH: 9d126f
ARCHIVEH: b8740c
Each data type can have its own hash and its own reconstruction cutoff. That is what makes the copy selective.
05 / Interactive endblock selector

Choose the cutoff for each data type.

An endblock value defines where reconstruction stops. Different data types can use different values, allowing the target environment to include exactly the required depth of history.

Move the controls to see how a selective restore changes its payload.

Restore recipe / target QA-07
82
55
21
158Selected blocks
47%Payload avoided
QA-07Target environment
Illustrative controls. The patent claims the selection mechanism; these values visualize the behavior rather than report a production benchmark.
06 / Incremental reconstruction

Refresh forward. Do not start over.

The target retains migrated portions of system data. A later restore can append the next relevant section instead of recopied history.

The system can also reconstruct a combined section from two previously selected sections. This supports iterative lower-environment assembly.

Target chain / second refresh
RESTORED 01RetainedH:a431
RESTORED 02RetainedH:bb72
RESTORED 03RetainedH:c091
INCREMENT 04AppendH:d173
FULL RECLONE

Repeat every byte, every refresh, including previously copied data.

SELECTIVE APPEND

Keep the useful reconstructed section and add the next relevant increment.

07 / Branch, test, rebroadcast

A lower environment can return a valid increment.

The patent also describes adding an incremental block in the target system, rebroadcasting its content to the source chain, and generating a new source-chain header.

Target-only increments can also be trimmed to restore the lower environment to its selected section.

Controlled branch behavior
SOURCE01 -> 02 -> 03
Parent blockchain in production
CLONE01 -> 02 -> 03 -> T1
Target adds an incremental block for testing
REBROADCASTcontent(T1) -> source -> 04
Source receives content and creates a new block header
TRIMtarget - T1
Remove target-only increments when restoring the selected state
08 / Practical applications

Where this pattern earns its keep.

The patent expressly identifies QA, testing, and development copies. The same architectural pattern suggests broader applications wherever large, verifiable data histories must be reconstructed selectively.

The examples at right are applications inferred from the claimed mechanics, not claims that these deployments exist.

Application landscape
ExpressQA environment refreshCopy only the data slices and history depth needed to reproduce a defect.
ExpressDevelopment sandboxesReconstruct a smaller, current environment without a full production clone.
InferredAudit reconstructionBuild an integrity-verifiable historical slice for investigation or controls testing.
InferredDisaster recovery tiersPrioritize the sections needed first, then append relevant history incrementally.
InferredData portabilityMove a qualified subset between permissioned environments with typed verification.
InferredRegulated test dataReduce unnecessary copying by reconstructing a narrower operational subset.
09 / Referenced by later patents

The idea travels across blockchain infrastructure.

Google Patents currently lists 19 citing publications for this granted patent, spanning fintech, enterprise infrastructure, and digital assets — including IBM, Hitachi, Alipay, Tencent, FMR LLC, Worldline, and WGC (UK).

These are forward patent citations. A citation shows relevance to later patent examination or disclosure; it does not by itself prove commercial use.

Forward citations shown by Google Patents / checked June 10, 2026
Asynchronous processing of blockchain blocksALIPAY / US11394584B2Published 2022
Cross-blockchain data migrationTENCENT / US20220229577A1Published 2022
Correcting historical records in distributed ledgersFMR LLC / US11522726B1Granted 2022
Digitize the value of a commodityWGC (UK) / WO2025056986A1Published 2025
Blockchain data storage optimizationIBM / US11195180B2Granted 2021
Test environment for blockchain systemsHITACHI / JP7266147B2Granted 2023
Digital payment infrastructureWORLDLINE / US11934386B2Granted 2024
Digital asset infrastructureNYDIG / US20230043702A1Published 2023
19 total forward citations. Additional citing entities include Truckl (supply chain), Hefei Dappworks, Mythical, Advanced New Technologies (Ant Group), Beijing University of Posts & Telecommunications, WRT Tech, and others. Google Patents also shows 14 related patent families.
10 / Patent record

A compact record. A durable architecture.

US10671315B2 frames blockchain as infrastructure for selective, integrity-aware system reconstruction. The lasting insight is architectural: make a large production history addressable, typed, and incrementally reusable.

Aug 2018
Priority Filed
US16/243,985 filed
priority date established
1 yr 6 mo
Feb 2020
Published
US20200057565A1
claims publicly visible
4 mo
Jun 2020
Granted
USPTO granted US10671315B2
18 claims, 3 independent
~18 years
Oct 2038
Expires
Adjusted expiration
per Google Patents
Read the granted patentView the applicationBrowse Siten Sanghvi patents
End / Patent 02