RuntimeSave: A Graph Database of Runtime Values
summary
Stores JVM variable snapshots into Neo4j graph database using a two-layer model with hashed+metadata nodes and Git-style content-addressing; implements graph-shape packers to collapse sparse arrays and common patterns, then computes per-node hashes and ID-hashes to deduplicate subgraphs. Prototype is an IntelliJ plugin that yields up to 50× node reduction, 20× fewer edges, and >10× faster insertion versus naive mapping, but full-line saving via JDI slows execution ~30–40×; future work targets bytecode instrumentation.
AI-written summary · about
chapters (10)
- 0:00RuntimeSave concept
- 1:07Collecting runtime values
- 3:17Graph model of runtime data
- 5:45Performance challenges
- 8:12Graph hashing algorithm
- 13:11Storing in Neo4j
- 14:34Performance improvements
- 16:20Applications and demo
- 17:25Conclusion and future work
- 18:09Q&A
related talks
30mPerformant GraphRAG Improving Neo4j to Drive Autonomous AI ...
49mModern Graph Databases
24mOur daily graphs written in Ruby and Neo4j
41mGraphGen: Conducting Graph Analytics over Relational Databases
26mHow NEO4J Saved my Relationship
28mKnowledge Graph of Your IT Landscape: Code Is Graph Data
50mThinking in Relationships: Practical Graph Database Modelling in Java with Neo4j
12mTEGRA: Efficient Ad-Hoc Analytics on Evolving Graphs