GraphOne: A Data Store for Real-time Analytics on Evolving Graphs

on YouTube ↗27m

" summary

GraphOne is a graph datastore that combines edge list and adjacency list storage with dual versioning and a GraphView API that lets analytics choose visibility between fast log writes and indexed snapshots. It sustains ingestion near 43.7M edges/sec before archiving, reports 2–3 orders of magnitude faster ingestion than graph databases, 5.36x update rate and >3x better analytics than a leading dynamic system, and runs within ~47 GB after cacheline allocation optimizations while offering snapshot isolation via MVCC.

AI-written summary · about

" chapters (11)

1. 0:00Intro
2. 0:20Graph Applications: Big Data and Scientific Applications
3. 1:45Definitions
4. 5:24GraphOne: Background and Architecture
5. 9:06Hybrid Store Details
6. 13:08Hybrid Store: Optimizing the Archiving Phase
7. 13:59Hybrid Store: Optimizing Memory Usage
8. 16:19Results: Hybrid Store Composition and Ingestion Rate
9. 18:08Results: Dynamic Graph Systems
10. 19:04Results: Cost of Data Management Over Static Graph Engine
11. 20:25GraphOne: Conclusion

" related talks

• 49mModern Graph DatabasesDavid Ostrovsky · Build Stuff 2025
• 30mAerospike Graph: Massively Scalable, High PerformanceVikranth Dharamshi · Aerospike APJ Summit 2023
• 35mBuilding Real-Time Graph Analytics PlatformsMuthuselvam Chandramohan · Conf42 Platform Engineering 2025
• 12mTEGRA: Efficient Ad-Hoc Analytics on Evolving GraphsIon Stoica, Joseph E. Gonzalez, Anand Padmanabha Iyer, Qifan Pu, Kishan Patel · NSDI ’21
• 29mAll about graphs in DatabaseMelliyal Annamalai · Oracle DatabaseWorld 2021
• 37mGraph Computing at ScaleMatthias Broecheler · Strange Loop 2013
• 44mHow We Store and Analyze a Large Social GraphMaksim Bartenev · HighLoad++ 2014
• 41mGraphGen: Conducting Graph Analytics over Relational DatabasesKonstantinos Xirogiannopoulos · PyData DC 2016