Multi-tenant Database Data Deletion via Graph Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-tenant databases, efficiently deleting data of former customers while maintaining compliance with data privacy and anti-money laundering regulations is complex due to interconnected document relationships, making it difficult to identify and isolate data for deletion.
Innovation Solution
A data deletion framework that organizes data as a graph, uses a graph search algorithm to generate smaller relationship graphs for former customers, and employs metadata and document graphs to identify and traverse relationships, ensuring accurate deletion of documents without affecting active tenants.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data deletion is performed in a multi-tenant database with interconnected documents, then data privacy compliance is achieved, but the deletion process becomes unreliable and incomplete due to complex document relationships
Solution Approach 1:
The patent segments the complex deletion task by introducing a deletion plan data structure that divides the deletion process into discrete, manageable steps. Each step targets specific document types and relationships, allowing the system to systematically handle interconnected documents without missing dependencies or causing unintended side effects
Solution Approach 2:
The patent performs preliminary actions by generating a complete deletion plan before executing any deletions. The plan identifies all documents to be deleted, all relationships to be updated, and all tenants affected, allowing the system to prepare and validate the entire deletion sequence before making any actual changes to the database
2Reliability
If all documents related to a former tenant are deleted to comply with data privacy regulations, then privacy compliance is achieved, but transactional records required for anti-money laundering regulations are incorrectly removed
Solution Approach 1:
The patent applies local quality by treating different document types and relationships with different deletion policies. The system identifies specific document relationships that require retention (such as transactional records for anti-money laundering) and applies preservation rules to those specific cases while allowing deletion of other tenant-specific documents
Solution Approach 2:
The patent introduces a deletion plan as an intermediary data structure that mediates between the requirement to delete tenant data and the requirement to preserve transactional records. The plan explicitly tracks which documents should be deleted, which relationships should be updated, and which data must be retained, allowing both regulatory requirements to be satisfied simultaneously
3Productivity
If a simple deletion approach is used for former tenant data, then the deletion process is simple and fast, but it fails to account for documents shared with other tenants and circular dependencies
Solution Approach 1:
The patent performs preliminary analysis to identify circular dependencies and shared document relationships before execution. The deletion plan is generated in advance to account for all these complexities, allowing the actual deletion phase to proceed quickly without needing to handle complex logic during execution
Solution Approach 2:
The patent introduces dynamic decision-making into the deletion process through the deletion plan, which adapts to the specific structure of document relationships. The system dynamically determines which documents can be deleted, which relationships need updating, and which tenants are affected, based on the actual data topology rather than following a rigid predetermined sequence
Data Source
AI summary
In an example embodiment, data in a multi-tenant database is organized as a graph representing the relationships among all documents and tenants. Specifically, each document is represented as a node in the graph and each tenant also is represented as a node. The relationships between the documents themselves, or between a document and a graph, are then represented as edges in the graph. A list of tenants whose data should be marked for deletion (e.g., former customers who no longer have a relationship with the entity maintaining the database) may be maintained. Periodically (e.g., once a week), this list of tenants and the relationship graph are fed as input into a graph search algorithm that generates smaller relationship graphs comprised only of documents associated with those accounts.