Fusion Computing at Ant Group
summary
Ant Group describes a production fusion computing platform built on Ray and Kubernetes that runs about 200k CPU cores and scales to ~1,000 nodes and 10,000 actors per cluster. It fuses streaming, graph, offline training and online decision services to cut data movement, achieve ~70,000 throughput/sec and low latency, and support dozens of core businesses. A payment-routing use case runs large-scale simulations and online optimization to minimize routing cost — a change-per-transaction impact measured in billions of dollars if not optimized.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 0:26Ant Group
- 1:16Fusion Engine
- 2:03Architecture
- 3:17Production
- 4:27Performance
- 5:17Payment Routing
- 8:48Customization
- 10:39Fusion Computing
related talks
10mKeynote: Building a Fusion Engine with Ray
26mUsing Ray On Large-scale Applications at Ant Group
02mStarcluster
35mBuilding a 5-Exaflop Supercomputer for Meta-AI Research and...
37mBuilding and operating large-scale streaming applications at King
35mApache Flink 2.0: Unified Enterprise Data Processing System and Beyond
31mBuilding a Virtual Biology Platform on Ray
21mKeynote: CERN Experiences - Ricardo Rocha & Clenimar Filemon