Backup Snapshot Vectors for Stable Enterprise RAG Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data management systems face challenges in efficiently utilizing backup data for retrieval augmented generation (RAG) due to the significant IT investment and disruption risks associated with integrating live enterprise data, which can adversely affect computing systems.
Innovation Solution
A data management system (DMS) extracts and organizes data from snapshots to generate vectors, storing them in a vector database accessible to communication applications, allowing retrieval augmented generation using backup data without disrupting the computing system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If live enterprise data is integrated for RAG, then LLM response accuracy is improved, but computing system stability deteriorates due to disruption risks
Solution Approach 1:
The patent creates vector representations (copies) of enterprise data stored in backup systems without accessing or disrupting the live computing system. These vector copies are stored in a vector database and used for RAG operations, allowing accurate LLM responses while maintaining system stability by working with replicated data rather than live data.
Solution Approach 2:
The patent introduces a vector database as an intermediary layer between the backup data and the LLM. This intermediary stores pre-generated vector representations of enterprise data, allowing the LLM to query and retrieve relevant information without directly accessing or affecting the live computing system, thus resolving the contradiction between accuracy and stability.
2Reliability
If backup data is used for RAG, then system disruption is avoided, but data accessibility deteriorates
Solution Approach 1:
The patent performs preliminary actions by pre-generating vector representations of backup data and storing them in a vector database before they are needed for RAG operations. This preliminary processing makes the data easily accessible and queryable by the LLM without requiring complex access procedures, thus improving ease of operation while maintaining system stability.
Solution Approach 2:
By creating and storing vector copies of backup data in a dedicated vector database, the patent makes the data highly accessible through efficient vector search operations. The copying process transforms the data into a format that is easy to query and retrieve, resolving the accessibility issue while using backup data that doesn't disrupt the live system.
3Device complexity
If vectors are generated from snapshots, then data organization is improved, but processing time increases
Solution Approach 1:
The patent applies preliminary action by generating vector representations of snapshot data and storing them in a vector database in advance. This pre-processing organizes the data into structured vector formats with associated metadata, making future queries efficient. The one-time processing cost is offset by the speed of subsequent retrievals, resolving the contradiction between organization and processing time.
Data Source
AI summary
Methods, apparatuses, and computer readable media are configured to perform operations comprising: obtaining, by a data management system (DMS), a first snapshot of a computing system; generating, by the DMS, one or more vectors based at least in part on data from the first snapshot, the one or more vectors representative of one or more respective portions of text within one or more files represented by the first snapshot; adding, by the DMS, the one or more vectors to a vector database along with metadata or a pointer to the metadata; storing, by the DMS, the one or more respective portions of text in a secondary storage environment; and adding, by the DMS to a mapping log, respective indications of mappings between the one or more vectors and the one or more respective portions of text.