Self-organization of data storage

Patent metadata

Field	Value
Publication number	US-12229166-B2
Application number	US-202318205646-A
Country	US
Kind code	B2
Filing date	Jun 5, 2023
Priority date	Jun 5, 2023
Publication date	Feb 18, 2025
Grant date	Feb 18, 2025

Abstract

Provided herein is a method of storing an incoming dataset in a data mesh. The method may include a plurality of steps. The steps may include associating a metadata tag with a classifying feature and a storage instruction in a (first) relational database. The steps may include scanning incoming datasets to identify datasets characterized by the classifying feature. The steps may include tagging an incoming dataset to generate a tagged dataset. The steps may include storing the tagged dataset in the data mesh, according to the storage instruction. The steps may include associating, in a second relational database, the metadata tag with the initial storage location. The steps may include modifying the storage instruction. The steps may include storing the incoming dataset an additional time in the data mesh, according to the modified storage instruction.

First claim

What is claimed is:

1. A method of tuning data storage within a data mesh based on machine learning outputs, said method using a processor, said method comprising the steps of:
associating in a first relational database, a metadata tag with:
a classifying feature; and
a first storage instruction;

identifying an initial storage location for an incoming dataset within the data mesh, the identifying being tunable using a machine learning system, the machine learning system executing on the processor, the tunability comprising:
upon receipt of said incoming dataset by said data mesh, ascertaining whether said incoming dataset comprises said classifying feature;
upon confirmation that the incoming dataset comprises said classifying feature, using the processor to:
obtain said metadata tag and said first storage instruction from said first relational database;
annotate said incoming dataset with said metadata tag, thereby generating a tagged dataset;
store said tagged dataset in the initial storage location, according to said first storage instruction; and
associate, in a second relational database, said metadata tag with said initial storage location;

subsequently modifying the first storage instruction using the machine learning system, the modifying comprising using the machine learning system to:
identify a previously stored dataset that is possibly a closest match to the incoming dataset;
determine a degree of match between the previously stored dataset and the incoming dataset based on a set of predetermined parameters;
derive storage rules from a storage instruction corresponding to the previously stored dataset in response to determining that the degree of match is the closest match; and
generate a modified storage instruction using the storage rules, the modified storage instruction comprising an instruction to store the incoming dataset in a second storage location that differs from the initial storage location;

connecting the initial storage location to the second storage location via a data link, the data link comprising a transmitter, a receiver and a data telecommunication circuit, the data link being governed by a link protocol; and
upon generation of said modified storage instruction, transferring, via the data link, said incoming dataset from the initial storage location to the second storage location.

2. The method of claim 1 , further comprising the step of automatically associating, in said second relational database, said metadata tag with said second storage location.

3. The method of claim 1 , wherein said initial storage location and said second storage location have different requirements for data storage.

4. The method of claim 1 , wherein said modified storage instruction reflects constraints of said second storage location that do not exist for said initial storage location.

5. The method of claim 1 , wherein said modified storage instruction reflects said second storage location not being subject to constraints that exist for said initial storage location.

6. The method of claim 5 , wherein said constraints comprise a limited storage lifespan for at least some data fields in said initial storage location.

7. The method of claim 6 , wherein said modified storage instruction is automatically triggered by an indication of said limited storage lifespan.

8. The method of claim 1 , wherein said modified storage instruction comprises denying user accessibility of at least one data field in said second storage location, wherein said at least one data field is accessible in said initial storage location.

9. The method of claim 1 , wherein said modified storage instruction comprises making accessible at least one data field in said second storage location, wherein said at least one data field is not accessible in said initial storage location.

10. The method of claim 1 , said method further comprising the step of automatically gatekeeping data transfers along said data link.

11. The method of claim 1 , wherein the steps of
(i) subsequently modifying said storage instruction; and
(ii) automatically further storing said incoming dataset;
are repeated a plurality of times.

12. The method of claim 1 , wherein said incoming dataset originated in a first jurisdiction, wherein said initial storage location is housed in said first jurisdiction; and wherein receipt of said incoming dataset in a (sub) network housed in a second jurisdiction automatically triggers said step of subsequently modifying said storage instruction.

13. The method of claim 12 , wherein said second storage location is housed in said second jurisdiction.

14. The method of claim 1 , wherein said storage instruction is derived from a storage instruction associated with a prior dataset annotated with said metadata tag.

15. The method of claim 14 , wherein said storage instruction is produced by a tunable storage rule engine.

16. The method of claim 1 , wherein said metadata tag is in binary format.

17. The method of claim 1 , said method further comprising the step of identifying redundant copies of said incoming dataset in said data mesh.

18. The method of claim 17 , said method further comprising the step of deleting said redundant copies of said incoming dataset.

Assignees

• Bank Of America

Inventors

• Meedinti Bhaskara Reddy Harinath
• Kurian Manu
• Varma Jayachandra
• Perkins Erica
• Solow Aeric

Classifications

• G06F16/285 Primary

  Clustering or classification · CPC title