RiftANN: Efficient Graph Traversal for Vector Search with RDMA-Based Memory Disaggregation

2026-10-06

AI Summary

This study addresses the challenge that billion-scale vector graph indices exceed single-machine memory capacity, leading to high search latency under RDMA-based disaggregated storage architectures. To this end, it proposes an approximate nearest neighbor search system built upon passive RDMA-disaggregated memory. The approach optimizes remote memory access by integrating compact vector navigation with exact verification. Furthermore, it introduces threshold-calibrated verification, bounded RDMA pipeline parallelism, and an impact gating mechanism to achieve safe and efficient graph traversal. Experimental evaluations on benchmark datasets such as SIFT demonstrate that the proposed system attains up to a 5.6× reduction in latency compared to DistVS and a 2.2× improvement in throughput over DiskANN.

Abstract

Graph indexes for billion-scale vector collections can exceed a single server's DRAM capacity. Passive RDMA-based disaggregated memory provides scalable capacity without memory-node computation, but conventional best-first search performs poorly in this setting. Fine-grained and unnecessary remote reads increase communication cost, while dependencies between candidate evaluation and subsequent expansion serialize traversal and leave CPUs idle. We present RiftANN, a graph-based approximate nearest neighbor search (ANNS) system for passive RDMA-based disaggregated memory. RiftANN navigates using compact vectors at compute nodes and selectively verifies exact vectors using a threshold calibrated from observed approximation errors. A bounded one-sided RDMA pipeline batches and overlaps the remaining remote accesses with local processing. RiftANN evaluates returned neighbors in parallel and incrementally incorporates completed results. An impact gate estimates whether unfinished evaluations may change the candidate set, allowing traversal to continue when their expected influence is small while preventing unsafe advancement from incomplete search state. A lightweight feedback controller coordinates RDMA concurrency with background evaluation capacity as search configurations and query loads change. We evaluate RiftANN on SIFT, DEEP, and SPACEV at 100M scale and SIFT1B. At matched recall, RiftANN achieves 1.6x to 4.3x latency speedups and 1.1x to 2.2x higher throughput than DistVS, and 1.6x to 5.6x latency speedups over SSD-based DiskANN and PipeANN. These results show that passive disaggregated memory can support low-latency graph-based vector retrieval without query-time computation at memory nodes.

Problem

• vector search
• graph traversal
• memory disaggregation
• RDMA
• approximate nearest neighbor

Innovation

• Approximate Nearest Neighbor Search
• Memory Disaggregation
• RDMA
• Graph Traversal
• Vector Search