CleANN: Efficient Full Dynamism in Graph-based Approximate Nearest Neighbor Search

2025-07-26

AI Summary

Existing graph-based Approximate Nearest Neighbor Search (ANNS) indexes are primarily designed for static datasets, struggling to simultaneously achieve high efficiency and accuracy under dynamic workloads involving insertions, deletions, and queries. Updates often degrade query quality, while graph maintenance typically requires expensive global reconnections. This paper introduces CleANN—the first fully dynamic graph index system supporting high-throughput concurrent insertions, deletions, and queries. Its core innovations include: (1) a workload-aware local linking mechanism; (2) query-adaptive neighborhood merging; and (3) semi-lazy memory reclamation. Evaluated on seven benchmark datasets, CleANN achieves query accuracy comparable to static rebuilt indexes, while delivering 7×–1200× higher concurrent throughput than state-of-the-art dynamic ANNS methods—marking the first solution to unify high accuracy with high throughput in dynamic graph-based ANNS.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Search and Optimization: Distributed Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Search and Retrieval-Augmented AI: Web crawling and indexing

Abstract

Approximate nearest neighbor search (ANNS) has become a quintessential algorithmic problem for various other foundational data tasks for AI workloads. Graph-based ANNS indexes have superb empirical trade-offs in indexing cost, query efficiency, and query approximation quality. Most existing graph-based indexes are designed for the static scenario, where there are no updates to the data after the index is constructed. However, full dynamism (insertions, deletions, and searches) is crucial to providing up-to-date responses in applications using vector databases. It is desirable that the index efficiently supports updates and search queries concurrently. Existing dynamic graph-based indexes suffer from at least one of the following problems: (1) the query quality degrades as updates happen; and (2) the graph structure updates used to maintain the index quality upon updates are global and thus expensive. To solve these problems, we propose the CleANN system which consists of three main components: (1) workload-aware linking of diverse search tree descendants to combat distribution shift; (2)query-adaptive on-the-fly neighborhood consolidation to efficiently handle deleted nodes; and (3) semi-lazy memory cleaning to clean up stale information in the data structure and reduce the work spent by the first two components. We evaluate CleANN on 7 diverse datasets on fully dynamic workloads and find that CleANN has query quality at least as good as if the index had been built statically using the corresponding data. In the in-memory setting using 56 hyper-threads, with all types of queries running concurrently, at the same recall level, CleANN achieves 7-1200x throughput improvement on million-scale real-world datasets. To the best of our knowledge, CleANN is the first concurrent ANNS index to achieve such efficiency while maintaining quality under full dynamism.

Problem

• Efficient dynamic updates in graph-based ANNS indexes
• Maintaining query quality during data insertions and deletions
• Reducing costly global graph structure updates

Innovation

• Workload-aware linking for distribution shift
• Query-adaptive neighborhood consolidation for deletions
• Semi-lazy memory cleaning for stale data