The Open-Source Distributed Graph Database: Nebula Graph
summary
Nebula Graph claims linear scalability across hundreds of billions of vertices and trillions of edges with many-second latency under high-concurrency OLTP workloads and community benchmarks from Tencent and Meituan showing top throughput. It uses RocksDB-style KV storage (transcript: roxdb), hash sharding, Raft for replication, duplicated edges for locality, and a separated query/meta/storage role architecture. v3.0 (Feb. ’22) adds KV separation and openCypher support; tooling includes Studio, Dashboard, a Kubernetes operator, Spark/Flink connectors and an SST side‑load path used by WeChat for t+1 ingestion.
AI-written summary · about
chapters (24)
- 0:00Nebula Graph intro
- 1:07Need for graph DBs
- 6:25What is a graph DB
- 7:26Why Nebula exists
- 9:32Nebula architecture
- 11:41Property graph model
- 16:46Storage engine design
- 23:54Query engine lifecycle
- 27:06Nebula Query Language
- 33:16OpenCypher in Nebula
- 33:58Another graph DB?
- 35:43Nebula ecosystem
- 37:09Studio demo
- 37:54Explorer features
- 39:01Nebula Graph Dashboard
- 39:42Cloud-native services
- 40:45Data import tools
- 41:51Exchange bulk loader
- 42:35Clients and connectors
- 42:57Graph analytics
- 44:02DB use cases
- 47:37Rail Search demo
- 51:30Amundsen lineage demo
- 56:09Recap and closing
related talks
49mModern Graph Databases
37mGraph Computing at Scale
01h 00mWho’s Afraid of Graphs?
30mAerospike Graph: Massively Scalable, High Performance
19mNuGraph: GraphDB as a Cloud Service Built Upon JanusGraph and FoundationDB
54mDistributed Graph DBMSs — The Future for Analytics on Big Data?
30mThe database trends that are transforming your database infrastructure forever
26mHow NEO4J Saved my Relationship