Chaos is a graph processing system for analytics on big graphs using small clusters.
Building on X-Stream
Chaos builds upon X-Stream, extending its functionality to multiple machines. The key insight is treating the aggregate storage of all machines as a single flat disk, employing work stealing for load distribution.
Programming Model
The system exposes a familiar scatter-gather-apply programming model that developers can use to express graph algorithms naturally.
Key Innovation
Rather than trying to optimize data locality (which is difficult with power-law graphs), Chaos embraces uniform random distribution and uses work stealing to handle load imbalance at runtime.
Learn More
For comprehensive technical details, see our SOSP 2015 paper.