Hierarchical Vector Indexing for Multi-Tenant Nearest Neighbor Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional vector databases face scalability issues in multi-tenant environments, requiring significant memory and incurring high costs, and do not efficiently support nearest neighbor queries without causing processing latency.
Innovation Solution
A hierarchical indexing system is implemented using a primary data storage unit and an intermediate data storage unit, with clustering and centroid-based organization, allowing for efficient creation and update of indexes without re-indexing frequently, and utilizing a quick-retrieval data storage unit for fast query responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional vector databases use significant memory to store vector data and indexes, then nearest neighbor queries can be served faster, but the system becomes expensive and cannot scale efficiently in multi-tenant environments
Solution Approach 1:
The patent divides the vector data and indexes into multiple shards distributed across different storage systems. Each shard contains a portion of the data, allowing the system to scale horizontally by adding more shards rather than requiring a single large memory structure. This segmentation enables cost-effective storage while maintaining query performance through parallel processing.
Solution Approach 2:
The patent introduces a hierarchical index structure with multiple levels (e.g., leaf level, intermediate level, root level) that organizes data across different dimensions of abstraction. This allows the system to query efficiently by navigating through hierarchical levels rather than scanning entire datasets, reducing the need for large memory allocations while maintaining fast query response.
2Speed
If conventional vector databases maintain indexes in memory for fast queries, then nearest neighbor search performance improves, but processing latency increases when data changes frequently require re-indexing
Solution Approach 1:
The patent implements a dynamic index management system where indexes are automatically updated when data changes occur. The system can rebuild indexes asynchronously in the background rather than blocking query operations, allowing the index structure to adapt to changing data while maintaining continuous query service. This dynamic approach reduces processing latency by avoiding full re-indexing operations during peak query times.
Solution Approach 2:
The patent pre-creates index structures at multiple hierarchical levels during data ingestion, so that when queries arrive, the index is already optimized for fast retrieval. The system also performs preliminary index maintenance operations during low-activity periods to prepare for future query workloads, reducing the impact of data changes on query performance.
3Device complexity
If conventional vector databases use a single storage system for all data, then the system structure is simple, but scalability and cost-effectiveness are limited
Solution Approach 1:
The patent creates a universal indexing system that can handle multiple data types and query patterns through a single hierarchical index structure. The same index mechanism serves both exact match queries and approximate nearest neighbor searches, as well as supporting multiple tenant environments. This multi-functional approach maintains relative structural simplicity while achieving high scalability.
Solution Approach 2:
The patent introduces an intermediary indexing layer between the raw vector data and query processing logic. This intermediate index structure acts as a mediator that translates diverse query requirements into efficient retrieval operations, allowing the system to scale to multiple tenants and data types without proportionally increasing system complexity. The intermediary layer abstracts the underlying storage details from the query processing layer.
Data Source
AI summary
Techniques for implementing a vector database in a multi-tenant environment are described. A system creates an index of a tenant that scales efficiently in a multi-tenant environment. The index is created by clustering the plurality of vectors into a set of clusters. The created index forms a hierarchical index including plurality of layers and is stored in a primary data storage unit. The system includes an intermediate data storage unit to store new vectors and to avoid re-indexing every time a new vector with an associated operation such as insert, update, and delete, is added. Further, the system provides reliable nearest neighbor vectors from the created index of the tenant. Read operation is performed over the quick-retrieval data, primary data, and intermediate to determine the nearest neighbor vectors.