I/O Optimizations for Graph-Based Disk-Resident Approximate Nearest Neighbor Search: A Design Space Exploration

TL;DR

This paper introduces an I/O-optimized framework for disk-based approximate nearest neighbor search, demonstrating significant throughput improvements through a systematic combination of techniques validated on multiple datasets.

Contribution

It proposes a novel I/O-first design space exploration with a page-level complexity model and a new system, OctopusANN, that outperforms existing solutions in disk-based ANN search.

Findings

OctopusANN achieves 4.1-37.9% higher throughput than Starling.

OctopusANN achieves 87.5-149.5% higher throughput than DiskANN.

Memory-resident navigation and dynamic width provide strong standalone gains.

Abstract

Approximate nearest neighbor (ANN) search on SSD-backed indexes is increasingly I/O-bound (I/O accounts for 70--90\% of query latency). We present an I/O-first framework for disk-based ANN that organizes techniques along three dimensions: memory layout, disk layout, and search algorithm. We introduce a page-level complexity model that explains how page locality and path length jointly determine page reads, and we validate the model empirically. Using consistent implementations across four public datasets, we quantify both single-factor effects and cross-dimensional synergies. We find that (i) memory-resident navigation and dynamic width provide the strongest standalone gains; (ii) page shuffle and page search are weak alone but complementary together; and (iii) a principled composition, OctopusANN, substantially reduces I/O and achieves 4.1--37.9\% higher throughput than the state-of-the-art system Starling and 87.5--149.5\% higher throughput than DiskANN at matched Recall@10=90\%. Finally, we distill actionable guidelines for selecting storage-centric or hybrid designs across diverse concurrency levels and accuracy constraints, advocating systematic composition rather than isolated tweaks when pushing the performance frontier of disk-based ANN.