Towards High-throughput and Low-latency Billion-scale Vector Search via CPU/GPU Collaborative Filtering and Re-ranking

Approximate nearest neighbor search (ANNS) has emerged as a crucial component of database and AI infrastructure. Ever-increasing vector datasets pose significant challenges in terms of performance, cost, and accuracy for ANNS services. None of modern ANNS systems can address these issues simultaneously. In this paper, we present Fusion-ANNS, a high-throughput, low-latency, cost-efficient, and high-accuracy ANNS system for billion-scale datasets using SSDs and only one entry-level GPU. The key idea of Fusion-ANNS lies in CPU/GPU collaborative filtering and re-ranking mechanisms, which significantly reduce I/O operations across CPUs, GPU, and SSDs to break through the I/O performance bottleneck. Specifically, we propose three novel designs: (1) multi-tiered indexing to avoid data swapping between CPUs and GPU, (2) heuristic re-ranking to eliminate unnecessary I/Os and computations while guaranteeing high accuracy, and (3) redundant-aware I/O deduplication to further improve I/O efficiency. We implement FusionANNS and compare it with the state-of-the-art SSD-based ANNS system-SPANN and GPU-accelerated in-memory ANNS system-RUMMY. Experimental results show that FusionANNS achieves 1) 9.4-13.1× higher query per second (QPS) and 5.7-8.8× higher cost efficiency compared with SPANN; 2) and 2-4.9× higher QPS and 2.3-6.8× higher cost efficiency compared with RUMMY, while guaranteeing low latency and high accuracy.