Building Real-Time Graph Analytics Platforms
summary
Claims a real time graph platform that hits subsecond queries for fraud and security use cases using Apache Kafka event streams and partitioning by vertex or edge and replication choices favoring eventual consistency. Shows concrete integrations: CDC pipelines, GraphQL fronting, Spark connectors, and options like Neo4j Amazon Neptune or in‑memory engines. Recommends full graph backups or event-sourced incremental backups, load testing with a synthetic graph generator, and explores GPU acceleration, ML stitching, and serverless deployments.
AI-written summary · about
chapters (14)
- 0:00Introduction and Speaker Background
- 0:30Why Adopt Graph Databases?
- 1:48Challenges in Building Graph Databases
- 3:55Limitations of Traditional Databases
- 6:57Core Architectural Patterns
- 10:30Horizontal Scaling Strategies
- 14:39Storage and Query Optimization
- 20:34Integration Patterns
- 23:29Monitoring and Observability
- 25:31Backup and Recovery
- 27:29Capacity Planning
- 29:53Architectural Trade-offs
- 32:19Future Directions for Graph Databases
- 33:54Conclusion and Final Thoughts
related talks
49mModern Graph Databases
32mBuilding Real-time Systems with Open Source Technologies
47mBuilding a Real-Time Analytics Database
30mLarge Scale Graph Solutions: Use-cases and Lessons Learnt
37mGraph Computing at Scale
30mAerospike Graph: Massively Scalable, High Performance
47mReal-time Fraud Detection Challenges and Solutions
08mBuilding a Real-time Data Future