This article explains how Xiaohongshu built the REDgraph graph database to tackle ultra‑large social network queries, compares graph databases with traditional relational databases, showcases a Gremlin example, and highlights the scalability and efficiency benefits of storing relationships as first‑class citizens.
Xiaohongshu's REDgraph graph database faced high latency for multi‑hop queries, so the storage team redesigned the query framework using MPP‑inspired distributed parallel execution, edge‑partitioning, operator forwarding, and caching, achieving over 50% latency reduction and making three‑hop queries viable for online services.
To meet Xiaohongshu’s strict P99 latency targets for multi‑hop graph queries, the REDgraph team built a distributed parallel execution framework that eliminates global barriers, pushes operators to storage nodes, uses routing operators and a NeighborCache, cutting three‑hop query latency by over 50 % and enabling production deployment.
This article presents the design, challenges, and performance‑focused optimizations of REDgraph, a large‑scale graph database used at Xiaohongshu, detailing its architecture, edge‑partitioning strategy, distributed parallel query implementation, and experimental results that demonstrate significant latency reductions for multi‑hop queries.
