The JanusGraph FoundationDB Storage Adapter
summary
JanusGraph gets a working FoundationDB storage adapter that maps adjacency-list vertex/edge rows into FoundationDB using the directory layer and tuple encoding, stores values as serialized byte arrays, and inherits distributed ACID so operations that produced "edges to nowhere" on Cassandra now raise a conflict on commit instead of corrupting the graph. Tests are adapted from the BerkeleyDB adapter, current limits include ~5s transaction window, and future work targets Redwood support, ID allocation simplification, collocated query routing, and built-in full‑text indexing.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 0:23Agenda
- 0:51Property Graphs
- 2:06IoT
- 2:35JanusGraph
- 3:48Gremlin
- 4:07JanusGraph Architecture
- 5:00FoundationDB Value Proposition
- 7:17Phantom Edge
- 9:33Building a new adapter
- 9:51Data model
- 10:51Map
- 12:00Berkeley DB Adapter
- 12:29FoundationDB Adapter
- 12:53Future Work
- 15:42Get Started
related talks
19mNuGraph: GraphDB as a Cloud Service Built Upon JanusGraph and FoundationDB
36mIntroduction to JanusGraph Database
36mIntroduction to JanusGraph Database
46mFoundationDB: The Best-Kept Secret of New Distributed Architectures!
18mFoundationDB Record Layer: Open Source Structured Storage on FoundationDB
50mBuilding the Next-Generation Backend on FoundationDB
46mScaling Identification Solutions Using Graph DB
29mImplementing a Graph Database based on PostSQL
